Methods for assigning a condition to a participant

[diatomsRcool]

We need to have a discussion about how or if to model conditions that were assigned to a participant based on something like an ECG or a survey or a medication list or something like that.

[diatomsRcool]

• Could do this via a relationship between a Condition and a Procedure: maybe have associated_procedure - won't always have Procedure metadata
• method by which it was diagnosed. Could be attribute of condition as diagnosis_method or something like that

Do we need to include a measure of confidence? Example - Did the patient have a heart attack according to EKG: Yes, Probably, Unlikely, No

Leaning toward doing this via diagnosis_method

Do we want to distinguish between a physician reading a procedure result to diagnose and a physician examining a patient to diagnose? Do we want to track if the physician made a diagnosis vs based on just a numeric results?

[diatomsRcool]

We need to make a list of what the possible permissible values would be.
Can we have a high-level list with an option to be more specific? A reasonable taxonomy? Do we make our own or use an existing? Does a reasonable one exist?

[stephanieshong]

If we are capturing the high level provenance already ( i.e. survey, EHR, Claims, labs, NLP, claim authorization) perhaps having a
provenance_data_note_detail may be sufficient?

[diatomsRcool]

List of possible values for diagnosis_method:

• based on medication usage to treat a condition
  -- regular bronchodilator or steroid user - not PRN
  -- PRN user of bronchodilator or steroids
• hospitalized for a condition (or treated in ER)
• seen a health provider for a condition (maybe combine with hospitalized)
• under medical care but not hospitalized for....
• did a medical professional say you had a condition (asking participant) some times it just says "have you been told you have X"
• do you have a condition (asking participant)
• clinical diagnostic impression (whose CDI? I assume the examiner.)
• FDI - first diagnostic impression
• First examiner opinion (is this the same as CDI?)
• Examiner's opinion
• Second Examiner opinion
• have you been told by a dr or hospitalized for X
• combination of treatment and measurement
• based on a laboratory measurement
  -- diabetes if glucose is > 200 mg/dL
  -- hypertension based on two blood pressure readings taken by a physician
  -- diabetes "lower cut point" for hemoglobin a1c and blood sugar
  -- myocardial infarction - major Q wave abnormalities
  -- myocardial infarction - ST-Segment Elevation
  -- DIABETES BASED ON ABNORMAL OGTT
  -- hypertension based on blood pressure
  -- Chronic kidney disease using eGFR only
• based on a piece of equipment or a procedure
  -- Electrocardiograph
  --- Left Ventricular Hypertrophy Criteria: R > 20mm in any limb lead
  --- Left Ventricular Hypertrophy Criteria: R > 11mm in AVL
  --- Left Ventricular Hypertrophy Criteria: R in lead I plus S in lead III >= 25mm
  --- LVH Criteria - : S-T depression (strain pattern)
  --- LEFT VENTRICULAR HYPERTROPHY DEF.-INVERTED T PLUS ANY VOLTAGE POSS.-VOLTAGE BUT FLAT T
  --- MORRIS P (DEPTH,DUR >=.04 MM-SEC) LVH CRITERIA
  --- LEFT AXIS DEVIATION LESS THAN OR EQUAL TO -30 DEGREES LVH CRITERIA
  --- and assorted others....
  --- Minnesota Code - specific value
  --- adjudicated electrocardiograms
  --- original machine coded ECGs
  -- X-ray
  -- abnormality noted by radiologist
  -- post exercise ECG
  -- venogram
  -- doppler ultrasound
  -- Impedance plethysmography
  -- Fibrinogen leg scan
  -- autopsy - coroner/medical examiner form
  -- Ventilation/perfusion V/Q scan
  -- angiography (>70% obstruction)
  -- stress test
  -- sonogram
  -- isotope scan
  -- VENTILATION-PERFUSION LUNG SCAN
• final diagnostic summary
• Rose Angina questionnaire
• derived from other questions specifically referenced in data set
• Gothenburg criteria
• Novacode xxxx by software
• Cornell definition
• Left ventricular hypertrophy, by Sokolow-Lyon
• Ankle Brachial Index
• hospital discharge information - discharge diagnosis
• Hachinski history of stroke
• chart review (MD review)
• based on symptoms
• based on procedures (angina based on history of cath or revas)
• COPD by BD test
• diabetes status ADA
• different definitions: COPD - MESA, GOLD, 2005 PFT, STAGE
• DIABETES MELLITUS BY 2003 ADA FASTING CRITERIA ALGORITHM
• HYPERTENSION BY JNC VI (1997) CRITERIA
• Left Ventricular Hypertrophy Index > 120
• Oxfordshire Classification
• TOAST Classification

Some are really specific (CARDIA)
Examples for PAD (also some for stroke)

• PAD BASED ON ULTRASOUND/ANGIOGRAM - OBSTRUCT/ULCERATED ILIAC ARTERIES OR BELOW
• PAD BASED ON ABSENCE OF PULSE BY DOPPLER
• PAD BASED ON POS EXERCISE TEST
• PAD BASED ON SURGERY, ANGIOPLASTY, OR THROMBOLYSIS
• PAD BASED ON AMPUTATION BECAUSE OF ISCHEMIA OR GANGRENE
• PAD BASED ON EXERTIOAL LEG PIN RELIEVED BY REST
• PAD BASED ON ANKLE-ARM SYSTOLIC BP RATIO <=0.8
• PAD BASED ON ULTRASOUND/ANGIOGRAM - ABNORMAL AORTIC ANEURYSM
• PAD BASED ON SURG/VASC PROCEDURE FOR ABDOMINAL AORTIC ANEURYSM

[diatomsRcool]

All of these fall under the following categories

• based on a measurement observation
• based on an expert opinion/exam
• based on a procedure having happened
• based on a questionnaire or other published or reusable set of criteria
• based on medications

[diatomsRcool]

Brian: i would call this condition_basis rather than diagnosis_method because not all conditions are diagnoses (e.g., they can be symptoms)

[diatomsRcool]

I'm wondering if this collides with condition_provenance. We've been using clinical diagnosis for conditions that have a method specified, but maybe we should have nothing listed if the condition is diagnosed during the visit as a part of the study.

[diatomsRcool]

we should wait to implement for after we get expert feedback