We're beginnning implementation of our Eclypsis CPOE in our Cancer Center (see here). Our system does not come with pre-built order sets, so we're in the process of building our own. We're trying to balance patient safety (no overdoses, missed meds, etc), efficiency (which I define as the fewest clicks by the ordering physician consonant with the safety goal), and teaching (making sure that our fellows and trainees understand chemotherapy regimens).
Before we initiatied the process, we moved to standardize our orders in the adult and pediatric practice. The orders use a standard form, at the top of which are the patient demographics and identifiers, the patients' height, weight and body surface area, a reference for the regimen being used, the Hasketh emetic risk, and any hold or dose adjustment parameters.
In the body of the order appears first the start date for the regimen, the premeds (typically antiemetics), any hydration steps, then the chemotherapy treatment regimen itself, followed by monitoring parameters, and any follow-up necessary. At the bottom are the authorizing signatures, with contact information, date and time.
An simple example for single agent rituxumab is shown below.
One complicating factor is that this is a weekly order -- rituximab is given on day 1,8,115, 22 -- not unlike many standard chemotherapy orders. My typical workflow is to write this order "drug by drug" when I see the patient, deposit it in a central file or repository where it is visible to the infusion area and the infusion pharmacy on each day that the patient is sheduled to appear. For safety and convenience reasons, I want to compute the chemotherapy dose once, and have this dose (also any reason for dose modification) propogate through to each day's order, pharmacy and nursing worklist.
Also, if the patient doesn't appear for a day 8, 15, etc dose, or more importantly, requires a dose modification, I want the modification to be calculated based on preset or "hold" parameters, or I want to be notified to make the dose adjustment myself.
Mulitple Day Orders
A more complex example of an order is shown below. This is taken from a pediatric acute lymphocytic leukemia (ALL) protocol. Our sample order starts on day 29 (January 4th) of the protocol.
Note that the example contains multiple days of chemotherpy, almost all of which are different from each other. However, the doses of each individual drug are the same, though the drugs may be given on different days. For example the vincristine is given at 1 mg -- the same dose -- on day 35 (Jan 18) and day 42 (Jan 25).
Translating into a computerized templated order
What is the best way to translate these orders into a computerized template? There are two possibilities, and I don't know which is the safest and most efficient. (Assume that the patient's height, weight and other variables are entered into the system, as well as any other information necessary to autocalculate the dose, and that the doses are automatically calculated and presented).
"Day by Day" or "Horizontal" ordering:
For the Pediatric ALL order, one could start with Day 29 ( that is, in our example Jan 4th), and be presented with, in succession, the opportunity to click the methotrexate order, the Cytoxan order, the Ara-C order and the pentamadine order. Next, one would be presented with Day 30 (e.g. Jan 5th), and click to confirm the Ara-C order (again with the dose precalculated). And so on for Day 31, 32, 33, day 35 (when again one would be presented with vincristine and E. asparginase), day 37, and so on.
I have called this the "horizontal method" as each day mimics a box on the conventional chemotherapy flow sheet which most people use. We have built the orderset in this horizontal style. (I hope to add a screenshot soon).
"Drug by Drug" or "Vertical" ordering:
In contrast, and perhaps illustrated by the simple rituximab example above, is the "drug by drug" or horizonatl method. Again, to mimic the flow sheet, one would move line by line, in the our ALL example, ordering first methotrexate, and have the system populate the appropriate days when methotrexate is to be given, then cytoxan, again with the system populating which days are appropriate, and on down through the Ara-C, pentamidien, vincristine, E. asparginase, etc.
I think that the drug by drug or "vertical" uses the fewest clicks and is most efficacious, and mimics my way of thinking. However I am open to suggestions or any knowledge from the literature.
A clinical trial experiment?
The best way to settle this would be an experiment -- that is build ordersets both ways -- randomize ordering physicians to either the "day by day" method or the "durg by drug" method and see which was safest and most efficacious.
Anna E. Schorer, MD
Jan, 25 2010 10:41 PM
The "mental model" most of us have for chemotherapy is that all drugs to be given sit together. For usability, the "big picture" should be visible. Furthermore, the clinician wants to have an order set ready to launch with the fewest possible modifications and "clicks" as possible.
A colleague who is an informatician has said that “technology is a ruthless enforcer of the rules”, while another colleague was fond of saying “technology is a light that makes black and white what formerly was gray”.
Here's a few reactions, based on our work implementing electronic chemo ordering in the VA:
Example 2 (supportive medications: the first major DHAP publication gave mannitol 50 gm IV Q4h prior to and during the infusional cisplatin. Lots of places no longer use mannitol at all for cisplatin and some more recent publications don't specify mannitol. These types of supportive oncology issues don't get the same intense interest as clinical trials for novel therapies, so our oncology community will need to dialogue (at the facility or at the national level) about reasonable "rules" for creating and adapting specific patient treatments from published clinical trials.