Sometimes, a Physician Is Not In Charge

Medical care is delivered in four distinct settings. They differ in many ways: who is in charge, what the immediate goal is, what information must be accessed or created, and what types of action are taken. The settings are: emergency/surgery/critical care; care on the regular hospital wards; diagnostic and consultative care; and ambulatory care.

Everyone is familiar with the critical care scenario. The doctor is in charge, commanding the situation and issuing orders. Practicing in this setting is a bit like flying a supersonic fighter 100 feet off the deck. The doctor (pilot) needs a heads-up display so that the critical variables can be monitored while watching the onrushing terrain. There is very little need to look back. To do so could result in a crash. The optimal data display in this setting is a spreadsheet format with lots of graphs. After the initial evaluation, which is similar to the ambulatory setting, the documentation is brief and procedure oriented.

Although orders are issued, an order-entry process is inappropriate. If equipment or a drug is needed, it is needed now. It should be drawn from inventory, administered and charted. As the University of Pittsburgh discovered a few years ago*, forcing a traditional order entry process into the ICU was killing people.

In the non-critical care settings many things are different, but none more so than that an "order" is rarely a command. On the ward, many doctors make requests related their patients. Not all of the requests are of equal importance and, usually, there are inadequate resources to fulfill all of them. The primary task is one of coordination and brokering. This is done by the nurses, especially the charge nurse. All of the requests must be periodically evaluated and re-prioritized based on available resources and patient condition. When services are required of the ancillary departments, the nurse's job is to marshal the necessary resources so that the patients get what they need. The type of data system that would be optimal to assist this sort of effort should probably look more like eBay than a typical hospital information system.

Such systems do not yet exist. The systems that do exist segregate the "orders" on individual patients so completely that the nurses are forced to invent their own mechanisms of performing the prioritization.

The diagnostic and consultative setting is perhaps the most compatible with EHR capabilities that are available today. The basic needs are to be able to access the patient's existing medical records and then add to them, often in a highly narrative fashion. The ability to include audio and video recordings in the record could greatly facilitate the work in many specialties as diverse as pathology and cardiology.

In the first three settings, a great deal of what gets recorded in the medical record is of no lasting value except for medico-legal protection (e.g. the I&O from the night shift.) All that really needs to end up in a patient's longitudinal medical record is a thorough summary of each ER visit, surgery, or hospital stay. Why the patient was treated, what the problems were, what significant events occurred that might have lasting implications for future treatment, etc. In short, a good discharge summary.

The final setting is ambulatory care. Here, as much as we might wish otherwise, the patient is in charge. The task is to find out if and how the patient followed any recommendations given at prior visits, evaluate the patient in that light and then make revised recommendations. This cannot be done effectively, especially in a group practice, without good records of what happened and what was recommended at the prior visits.

In this setting, "orders" are not orders; they are advice to the patient. The patient will ultimately choose when, where, and if to get prescriptions filled, take the medications. They will decide to follow the diet and exercise recommendations, or not. In this setting the practitioner's role is that of salesman and coach, not commander.

There are two take-home messages:
1. Recognize and accept that the physician is not always in charge.
2. An EHR that provides only a single data display, work flow, and ordering approach will probably be unsuitable for at least two, and perhaps all, of these settings. There are good reasons to use a different EHR in each setting that is optimized for it. Appropriate summaries should be sent to the longitudinal record so that a functionally complete record exists in a single location.

*Han YY, et.al.; Pediatrics. 2005 Dec;116(6):1506-12; Unexpected increased mortality after implementation of a commercially sold computerized physician order entry system.; Department of Critical Care Medicine, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA.

For more about Dan Essin and our other Practice Notes bloggers, click here.