drbowen wrote on Tuesday, December 06, 2005:
Decision support will take more thought and effort.
I have approached this so far by designing forms that are driven by chief complaint.
I start with a chief complaint such as "bronchitis".
I then research a general way of taking care of “bronchitis”. Currently I use “Saunder’s Manual of Medical Practice”. This book is very general and oriented towards family practice and urgent care.
I then create the form using the primary diagnostic clues as stated in “Saunder’s Manual of Medical Practice”, include the differential diagnoses, “clinical pearls” that help delineate different conditions, and a short list of diagnoses in a drop down box.
Then I go through the process of creating an OpenEMR "form". I think these might be more appropriately referred to as "clinical modules".
I would prefer to incorporate the search engines that you use in your "fee sheet". I could look up the diagnosis codes and fee codes from inside the "clinical modules". The billing function could then be initiated from inside of the encounter using the "clinical module" and the fee sheet search tools that you have already developed.
Your addition of the ability to associate a chronic diagnosis with a particular encounter is one of the first steps in this decision making process.
It does require standardization of the diagnoses being used. I think you should "require" the use of the fee sheet diagnosis search tool. None of this will work if the business logic can not make a match due to poor spelling or typos.
Say, as an example, I am seeing a 74 year old man with type II diabetes that is out of control.
I look up:
diabetes mellitus : 250.02 (poorly controlled type II DM)
I enter this as a chronic diagnosis.
I start an encounter using this diagnosis.
at the end of an encounter as I "finalize the visit". I get a pop up box that reminds me:
The following parameters needs to be checked:
*Hemoglobin A1c* (type II DM) quarterly
*Lipid panel* (Type II DM) anually and quarterly if abnormal
*Liver function* quarterly if on (search the med list for HmgCoA reductase inhibitors, thioglitazone (Avandia, Actos), etc.)
*renal function* (Type II DM) quarterly, warn if the last creatinine is/was 1.4 (men) or more and if 1.3 (women) or more. Critical warning if the medication list also shows any medication containing metformin.
*retinal exam* due every 6 months
*microalbumin to creatinine ratio* due annually or quarterly if elevated
*PSA* male over 50
One can see how ths will require some work. Ideally, a laboratory information system will become part of OpenEMR and these searches based on certain lab values will be checked by the "Decision support module" and reported only if data is missing.
In the "bronchitis module" "expert decision models" can be incorporated. These "Expert Decision Models" can arrive at the correct diagnosis with high degrees of certainty. This certainty level can be expressed as a probability.
This depends on setting up the forms correctly Signs and symptoms can "checked off" with explanatory text boxes.
The "Expert program would run off the presence or absence of the signs and symptoms (simple check boxes will do).
A “help” button to explain to the physician how the decision is being made and why. (Doctors don’t respond well to being kept in the dark.)
This type of decision making is why I have been setting up the "clinical modules" the way I do.
Sam Bowen, MD
