• The patient and physician are people. And they are both experts. Like other experts they may or may not be able to articulate their expertise autonomously.
  

• The patient and physician are different people with potentially the same or different concepts in mind when they discuss medical issues related to decision.

1) Elicit the speaker's terms and ...
2) with them generate a temporary understanding of the terms with their concomittant counterparts.
3) Use these constructs to ask questions that enable you to understand the relationships between the constructs.
4) capture those relations (temporarily) and use them to ...
5) test your understanding of the speaker by using the constructs and relations in a fashion that confirms your understanding.

The Larger Context for this Ambition:

What is not my goal:

Subgoals:

identify objects in patient's universe of discourse
identify the terms used in the patient's Interpretation
identify relations in patient's Conceptualization
identify objects in physician's universe of discourse
identify the terms used in the physician's Interpretation
identify relations in physician's Conceptualization
identify common conceptualizations that have different names employed to describe objects and/or relations
identify inconsistencies and incompleteness in need of resolution (medical consequences are highly likely to expose incompleteness in the patient's anticipatory capacities)
communicate findings to physician - Herbert Stoylan's formalizations?
communicate findings to patient - patient as an expert => Herbert Stoylan's formalizations?
resolve deficiencies and differences

Speculation based upon observation

Computer user parallel: Until a person is confident that the machine can perform as they currently do, they are not inclined to trust it to do anything else.

This is the proposed graphic representation of a generic bipolar construct (constellatory to healthy/unheathy) with a generic subordinate construct for each pole.

There would be an applet here if your browser supported java applets

Software

Omnigrid
is an example of current Repertory grid software.

• current analytical measures - uncertainty, salience, element distance.
  
• Scalar constructs allow correlation indices like Pearson correlation coef.

Enumerating the concievable relations thus far

• implication - four forms or subtypes
  Hinkle's four quadrants:
  Parallel
  Reciprocal
  Orthogonal
  Implicative Dilemma
  
• subsets: subordinate, superordinate and constellatory
  Venn diagrams
  subsumes not same as implication
  
• cause/effect
  
• examples: trees
  
• marked/unmarked
  
• positive/negative
  
• implication: four forms or subtypes
  Golden Section Ratio hypothesis
  
• nominative
  
• prototypic constructs
  
• three axis of trait structure evaluation (good/bad), potency (strong/weak), activity (passive/active)
  OF PARTICULAR INTEREST TO THE MEDICAL DOMAIN
  
• things over which patient has control vs no control, eg. Youth/older (A FORM OF CAUSE vs. AN EFFECT OF POOR HEALTH)
  

Problems with current data

How to terminate interview? Exhaustive approach so far

Dynamic list of constructs

• Fixed list in classical Repertory Grid
  
• could add to list as necessary
  
• Use large oversized list with instructions to only use those necessary
  
• Additionally at interview's end, could query adequacy of list and let this addition be selectively utilized
  - presumption: no one interviewed previously would have used this addtion else they would have felt the need to add it

What are and what can we do with results?

Compare outcomes of recent decision analysis trees from CDM [NEMC] consults on temporal arteritis and Hormone Replacement Therapy (HRT) risk analysis with a constructus image.

Recent Temporal Arteritis decision tree consultation at NEMC:
Options were no treatment, steroidal Rx w/ risk of cardiac event, rebiopsy
Outcomes were survival or death, survival further subdivided into monocular vision loss and binocular vision loss.
Could ask: "How does monocular vision loss effect each item viewed important by the patient by the constructs elicited? How does binocular vision loss effect each item viewed important by the patient by the constructs elicited? How does survival impact each? How does death impact each?"

HRT - list of outcomes considered in the impact of progestin/estrogen therapy were limited to cancer risk (breast and endometrial), coronary heart disease and osteoporosis. The units of comparison were QALE (Quality Adjusted Life Expectancy) years. Women with all but the highest number of risk factors for breast cancer with no risk factors for coronary heart disease would be the only ones who could not expect an extended life expectancy. Limitations sited included the weakness of quality of life considerations.

These conclusions could be made more relevant to the individual by considering the interaction of any or all elements of the patient's constructs with the either the diseases or the medication considered. The space of potential ramifications would be a subset of the cartesian product of (diseases & medications X construct elements). What does each symptom say about each item of importance revealed by the constructus? What does breast or endometrial cancer, and potential reconstruction say about each item of importance revealed by the constructus? What does reproductive tract cancer, resection, and potential reconstruction say about each item of importance revealed by the constructus? How might hormone therapy effect the things important to the patient? How will things important to the patient affect the efficacy of the treatment so as to assure the predictive fit of this analysis. This is admittedly an extension of the study's complexity that should be a study in its own right, however, the point here is relevance. The opportunity to increase the granuarity of consideration by a finite number of pertinent parameters exists as a result of acquiring these constructs by which the patient discriminates healthy from unhealthy.

Samples from 42 pilot "street" interviews that exemplify specified notations: