Phenotype Description: Patients on statins for primary prevention who develop an AMI or 1st AMI.
Below are algorithms used to identify AMI and 1st AMI cohort at BioVU. If you have questions regarding any of the information presented on this page, you may contact either:
Wei-Qi Wei at email@example.com
Joshua Denny at firstname.lastname@example.org
List on the Collaboration Phenotypes List
Type of Phenotype:
Friday, June 7, 2013
Owner Phenotyping Groups:
View Phenotyping Groups:
People Watching For Updates:
Wei-Qi Wei. Vanderbilt University. Statins and MACE. PheKB; 2013 Available from: https://phekb.org/phenotype/170
Troponin I labs
In the AMI algorithm you require confirmed labs Troponin I >= 0.05 or CK-MB/CK ratio >3.0 for case confirmation. Also requiring No abnormal TnI/CKMB ratio labs for control confirmation. Our Troponin I labs report results in text format of "<0.1" or "<=0.1" or "0.3" ng/mL in roughly 70% of the records returned for our PMRP cohort.
We have Troponin I labs as far back as December 1995 and they indicate "High" values based on the following timeframe:
High 0.6 ng/mL - Dec 1995 to Mar 5, 1998
High 0.3 ng/mL - Mar 6, 1998 to Mar 30, 2006
High 0.1 ng/mL - Mar 31, 2006 to Current
Question: I am unable to determine if "<=0.1" (text value) is >=0.05 (numeric), so can I just use our HIGH cut-offs based on the lab date and determine if a Troponin I lab result exceeds the HIGH cut-off (example: IF HIGH = 0.1 and result = "<=0.1" THEN not elevated)?
Yes using those cutoffs is fine.
Statin exposure reads:
The number of days that the earliest statin prescription prior to the event (AMI/Revascularization)
I'm assuming this should be
The number of days from earliest prescription of (any) statin to the event (AMI/Revascularization).
Is that correct? If not, please clarify your intent.
That is correct.
Also the way I am reading
Also the way I am reading your control definition there will be no controls
*No MACE found in problem list by NLP
*Matched to cases on MACE
These three seem to exclude everyone. If someone is on a statin and has no notes indicating prior MACE and then they have a MACE (so they can be matched to another case) aren't they then a case themselves?
What am I missing?
Apropos this question - there
Apropos this question - there are also four different types of cases, do you want controls equally distributed between being matched to AMI, 1st AMI, revascularization, and 1st revascularization or some other scheme?
Please include all the possible controls and cases, and don't worry about matching.
Can you clarify in the main
Can you clarify in the main case control logic, which CPTs are revascularization codes and which ones are specifically stent/angioplasty codes?
Controls and CPTs
Sorry for the confusion. Cases are subjects with MACE while on statins, and controls are subject with no MACE while on statins. This will be updated in the updated algorithm.
As to the codes:
33533 – coronary artery bypass, using arterial grafts; single arterial graft
33534 - coronary artery bypass, using arterial grafts; two coronary artery grafts
33535 - coronary artery bypass, using arterial grafts; three coronary artery grafts
33536 - coronary artery bypass, using arterial grafts; four or more coronary artery grafts
33510 - coronary artery bypass, vein only, single coronary venous graft
33511- coronary artery bypass, vein only, two coronary venous grafts
33512- coronary artery bypass, vein only, three coronary venous grafts
33513- coronary artery bypass, vein only, four coronary venous grafts
33514- coronary artery bypass, vein only, five coronary venous grafts
33515- coronary artery bypass (old code)
33516- coronary artery bypass, vein only, six coronary venous grafts
33517 - coronary artery bypass, using venous grafts and arterial grafts, single vein graft
33518 - coronary artery bypass, using venous grafts and arterial grafts, two venous grafts
33519- coronary artery bypass, using venous grafts and arterial grafts, three venous grafts
33520 - coronary artery bypass (old code)
33521 - coronary artery bypass, using venous grafts and arterial grafts, four venous grafts
33522 - coronary artery bypass, using venous grafts and arterial grafts, five venous grafts
33523 - coronary artery bypass, using venous grafts and arterial grafts, six venous grafts
92980 – transcatheter placement of an intracoronary stent, percutaneous, with or without other therapeutic intervention, any method, single vessel
92981 - transcatheter placement of an intracoronary stent, percutaneous, with or without other therapeutic intervention, any method, each additional vessel
92982 – percutaneous transluminal coronary balloon angioplasty, single vessel
92984 – percutaneous transluminal coronary balloon angioplasty, each additional vessel
92995 - percutaneous transluminal coronary atherectomy, by mechanical or other method, with or without balloon angioplasty, single vessel
92996 - percutaneous transluminal coronary atherectomy, by mechanical or other method, with or without balloon angioplasty, each additional vessel
C1874 - Stent, Coated/Covered, With Delivery System
C1875 - Stent, Coated/Covered, Without Delivery System
C1876 - Stent, Non-coated/Non-covered With Delivery System
C1877 - Stent, Non-coated/Non-covered Without Delivery System
Hi again (I'm on a roll!)
For the case defintion NLP I see the list of terms for the controls, but I don't see which terms you want us to look for for AMI and which are for revascularization. The control list seems to have terms that are not apropos either such as chyper so I didin't want to assume.
See updated algorithm for additions.
In the dictionary
*There is no definition given for the 0/1/2/3 and 0/1/2/3/4 MACE variables
*For diabetes do you want type 1/2/others?
*Which HTN meds do you want for your HTN def? If you asked me to define this I would grab ace inhibitors, beta blockers, ccbs and a grab bag of Clonidine, methyldopa, hydrochlorothiazide, fursemide, bumetanide, hydralazine, labetalol, terazosin, doxazosin, prazosin, acetazolamide, losartan and spronolactone.
*Smoking will also have a level of missing.
*We will not have aspirin use captured in any sort of reliable way since it is OTC.
*'Antiplatelet' is too broad for me. What are these drugs and what 'types' do they fall into?
I am updating the dictionary, to be posted.
*There is no definition given for the 0/1/2/3 and 0/1/2/3/4 MACE variables? This has been updated and added.
*For diabetes do you want type 1/2/others? T2DM
*Which HTN meds do you want for your HTN def? If you asked me to define this I would grab ace inhibitors, beta blockers, ccbs and a grab bag of Clonidine, methyldopa, hydrochlorothiazide, fursemide, bumetanide, hydralazine, labetalol, terazosin, doxazosin, prazosin, acetazolamide, losartan and spronolactone. Medications List added
*Smoking will also have a level of missing. Mark as Unknown if missing
*We will not have aspirin use captured in any sort of reliable way since it is OTC. We made this Not Required.
*'Antiplatelet' is too broad for me. What are these drugs and what 'types' do they fall into? Definition added.
1. Should we provide "all" eligible controls so that matching to cases can be done at VU? Depending on control counts, you may be able to match 2:1, 3:1, 4:1. If so, would it make sense to change date fields to age fields so that you can match by age (i.e. age_first_mention_statin, age_event (for MACE events), and age_most_recent_statin)?
I believe you stated (in a recent email) that controls are required to have >=180 days of Statin exposure prior to their most recent visit.
2. Do controls need to be seen recently? If so, how recently?
3. Are there any other age timepoints you would need for matching, such as age_at_most_recent_visit?
1./3. Go head and do not worry about control matching. Just provide all available cases and controls.
For 2, we do not need any requirement that they have been seen recently. Only that they have no MACE events between statins and their last visit.
"continuous" &/or "current" statin usage required?
What if >365 days btwn last statin Rx & last visit (& no MACE of course) -- can they really still be a control, as they may have stopped taking the statin (one would need a new Rx at least every 365d as MDs cannot prescribe infinite refills at least to my knowledge)? We have at least half of our (~1400) possible controls in this situation. Related question for cases:Do cases need to have their last statin Rx be within 365 days of the MACE? i.e., pts can stop taking statins for various reasons,and maybe they were not on a statin when the MACE occurred (although they were before)We have 2 possible cases like this -- there last statin Rx was >365 days before their MACE,so they were theoretically not on a statin when the MACE occurred. Thanks!
Thanks for all your answers!
Is the current data dictionary the final version, or is it still subject to change?
Please help me to understand following data dictionary variables
2. How_event_recorded: Values are 0-Code and 1-NLP: NLP is used for exclusion only so in which situation How_event_recorded=”y”? Also Mace exclusion NLP terms are same as case NLP terms.
3. AS per MACE algorithm there should be 2 ICD dx within 5 days window. I would suggest if we could make site specific modification as follow
a. For inpatient dx 5 days window not required but it should be Primary diagnosis for inpatient stay + elevated Troponin /CK MB lab within 5 days and Statin within 180 days b. For outpatient dx 5 day window rule applies.
Else we will be dropping out some genuine cases. E.g. An MI patient has inpatient stay for 7 days. In this situation we will not find 2 MI dx within 5 days. I agree with this modification.
I will let you know about 1 &
I will let you know about 1 & 2. But for now, your alteration described in 3 is ok.
For 1, If a subject has both
For 1, If a subject has both AMI and CABG, we will consider the subject as AMI. I think it is fine as a repeat variable since we will to the analysis on AMI cases and then release the dataset to cases with either AMI or CABG. We have removed Death from MACE Type as we do not want to include that in that covariate. If subject has died, include the date of death along with the MACE status.
For 2, that is correct. Yes, How_event_recorded would always be codes. NLP is used only in exclusion criteria.
A little more clarification on 1?
I am still a little unclear on how to handle cases that have more than one type of MACE. You say if the subject has both AMI and CABG, to have Mace_type=AMI. Following the same logic, if a patient has both AMI and PTCA, should we also call this Mace_type=AMI. Or, should we include both records (as you say it is fine as a repeat variable)?
MACE_Type as a repeat variable
Go ahead and send Mace_Status as a repeat variable.
More than one statin prescribed on same day?
I have 7 patients that were prescribed two statin meds on the same day/time (not simply generic vs trade name). How should I report the type of statin in these cases?
Include both separated by a comma.
I have one last question regarding the medications in the data dictionary. You have in the variable descriptions: "if the patient was using Asprin 180 days before the first event." Does this mean within 180 days before the first event? In which case, if the patient was on them, say 100 days before the first event, they would be flagged "yes" here.
Just realized that I never received an answer to this question, and now I have a follow up question as well...
How did your group handle start and end dates for medications? For example, if a patient was prescribed a med at a particular order date but has no record of an end date, would you condsider him to still be on the med, and for how long (six months, one year, indefinitely, etc...).
The medication part of the data dictionary is the last outstanding piece we have before we can complete the phenotype and send to you.
we just requested the first mention of statin in EMR and the most recent statin mentioned 180 days prior to the event.
As to your example, yes, we still consider the patient as taking statin.