An emergency department in a hospital provides an important example of a partially self-contained system that must respond effectively to a wide variety of circumstances due to the variety of individual problems, combinations of problems, and unscheduled demand, which fluctuates significantly over time. For a number of years, Mark Smith, MD, Chairman, and Craig Feied, MD, Director of Informatics of the Department of Emergency Medicine, Washington Hospital, Washington, D.C., have been developing strategies for applying complex systems ideas to the design of processes, information flow, and physical layout in an emergency department.
A specific example of applying complex systems ideas to the emergency department is the redesign of the process of obtaining blood tests. The process of blood testing that originally existed in the emergency department involved sending the blood samples to a central laboratory. The basic problem in this process was not just the distance over which the sample and test order had to be sent, but the number of people and communication steps that were involved. Specifically, the process involved eight steps.
The order for the test is written on the chart, where it is then noted by the nurse, and entered as an order into the "order entry" computer system by either the nurse or the clerk. A lab request slip is then generated on a nearby printer. The patient’s blood specimen is then attached to this slip and the pair sent, via pneumatic tube or transporter, to the "accessioning" position of the laboratory. The sample is then distributed to the section of the laboratory that is appropriate for that particular test. Once completed, the test results are transmitted back to the ED, via telephone call or fax to the unit secretary, who then must relay the results back to the physician who ordered the test. This process involves at least 7 people, 8 steps, and takes 60 minutes.
In an effort to reduce patient throughput time, a "point of service" testing system was instituted, using a "mini-laboratory" situated in the middle of the emergency department and staffed with a laboratory technologist:
To obtain a blood test in this system, the physician marks the desired test on a sheet of paper and hands that paper directly to the point-of-service laboratory technologist. The technologist locates the blood sample that had already been placed onto a rack by the patient’s nurse and then runs the test. Three minutes later, the technologist returns the sheet of paper complete with the test results to the physician; three people, three steps.
This change reduced the time it takes for a laboratory result to be completed and also reduces the potential for error of a common and central process in the emergency room. When it is possible, complex systems, including organizations, work better when the processes are designed with fewer steps and fewer people. If processes are kept short and simple, fewer opportunities exist for something to go wrong. Less degradation of information occurs, because there are fewer 'handoffs' of information from one person to another. Fewer steps mean that fewer feedback loops need to be constructed in order to ensure that errors or faults in the process are detected and corrected.
Just as a few simple rules and a small number of variables can create a complex adaptive system, small reductions in the number of variables and rules can have dramatic and nonlinear impact in reducing the complexity of a system. In the more complicated version of the lab process, the lab result time-to-physician was non-deterministic because of the complexity of the process. Results reporting sometimes took many hours; tests were sometimes lost completely and had to be repeated; and people sometimes simply forgot to see the results at all. Under the point-of-care process, the lab results time-to-physician is highly predictable and the error rate in results delivery is very low.
When changes are made, unintended side effects can occur that may be beneficial or detrimental. An unintended and unanticipated positive side effect occurred with the institution of point of service laboratory testing: the turnaround time for critical tests that were being sent to the main laboratory decreased. The reason was that the emergency department now had a laboratory technologist in its midst who in effect became an advocate for its interests. Whenever a test whose result was urgently needed had to be sent to the main laboratory for processing, the point-of -service laboratory technologist would convey the urgency of obtaining the result to her colleague in the main laboratory. That colleague responded to a call from a fellow laboratory technologist in a manner different from a call from a nameless and faceless doctor, nurse, or clerk.