Good Performance Support is a Model of Behavior
By Sam S. Adkins

Precision performance support requires high-fidelity behavior models.

Performance support products are designed to help people quickly find the information they need to complete tasks while they’re doing those tasks. The most effective performance support maps to very specific behavior. The better that behavior is modeled, the more precise the performance support becomes.

There are several situations in which the impact of precision performance support is profound, most notably in public safety, aviation, and healthcare. In these situations, precision performance support means saving lives in real time.

Saving lives in real time

Some of the most dramatic effects of performance support have been achieved in aviation. While commercial airline accidents are extremely rare, two-thirds of accidents are caused by pilot error. In 2004, NASA’s Aviation Safety and Security Program selected five human behavior modeling teams to develop human performance models (HPM) of pilots performing taxi operations and runway instrument approaches.

In a 2005 paper called ”Human Performance Models of Pilot Behavior” describing the results of the project, the authors concluded that, “Across all of the modeling efforts, the tools were able to predict errors that occurred because of situation awareness degradation, memory degradation and interference, airport layout, pilot expectation and habit, distraction, and workload.”

The positive impact of precision performance support is also evident in decision-support systems designed for healthcare professionals. The U.S. Food and Drug Administration (FDA) estimates that as many as 372,000 Preventable Adverse Drug Events (PADEs), occur each year. A study by the Institute of Medicine of 1,116 hospitals suggested that more than 44,000 deaths occur each year as a result of in-hospital medication errors. FDA studies have shown that PADEs occur at an average rate of 1.53 times per 100 admissions.

The FDA believes these errors are caused by administering incorrect dosages, errors in transcription, errors in filling prescriptions, and adverse drug interactions. It is estimated that there is approximately 40 percent error rate during drug interaction risk analysis, prescription, transcription, and dispensing.

Performance support can dramatically reduce this error rate.

For example, The Children’s Hospital of Pittsburgh launched a computerized physician order entry (CPOE) system called Children’s Net in October 2002. The CPOE has a warning system that provides an alert if a dose seems out of line for a particular child based on predetermined dosage-per-weight standards. The performance support has reduced medication errors by 75 percent and has virtually eliminated weight-related PADEs.

Mobile performance is mobile learning

Handheld clinical decision support is the most widespread type of performance support used in healthcare. Over one million healthcare professionals in the United States currently use handheld devices in their daily routines. The devices contain drug information, interaction decision trees, treatment strategies, specialty guidelines, and evidence-based medical information.

In a late 2005 survey conducted by Skyscape, a company that sells mobile performance support products, 84 percent of the 2,800 medical professionals surveyed believed they achieved a significant decrease in medical errors by using handheld clinical decision support.

For instance, nurses at St. Clair Hospital in Pittsburgh use Pocket PCs with clinical decision-support software called VeriScan to protect patients from medication errors. The devices are used to scan barcodes and RFID chips on patients and prescriptions. The VeriScan software confirms that a nurse has the correct patient, medication, time, and dose each time a medication is administered. The system has been in use for two years and has prevented more than 5,000 medication errors per year.

In another example, first responders and emergency medical technicians (EMTs) use a type of performance support called “emergency medical decision support,” and like the healthcare industry, these are increasingly handheld products. The Georgia Tech Research Institute (GTRI) developed a performance support product called "Chemical Companion” with funding from the federal Technical Support Work Group. GTRI provides the product for free to first responder agencies throughout the United States. The software contains detailed information on 130 of the most common chemicals associated with hazardous materials (hazmat) situations.

The first responders use the tool to determine what protective clothing is required, what equipment is needed, the chemical reactivity of particular hazardous materials, the distances required to establish protective zones, and the appropriate medical aid that is needed if people are contaminated. The tool also is used to determine hazardous material based on patient symptoms. If a first responder types in such patient symptoms as twitching, constricted pupils, excessive sweating, and confusion, they are advised that the patient is probably suffering from the effects of the nerve gas Sarin.

Over 20 first responder organizations in the state of Washington, including hospitals, fire and police departments, and decontamination units, use a handheld product from Iomedix called MobileIRIS (Incident Response Information System.) EMTs use the system to take the patient’s photo, record the patient's vital signs and injuries, and log treatment being provided at the scene. The data is beamed to participating hospitals and emergency operations centers. EMTs receive a range of performance support, including real-time coaching from hospital staff.

Ad hoc behavior modeling: Is it a simulation or is it real?

There is a new defibrillator device from Philips that “talks paramedics through CPR and shuts up when they get it right."  Phillips has a portable version of the products called HeartStart that is designed for home users. “When you turn on the HeartStart Home Debrillator, it starts talking to you, to guide you through each step.” A blue flashing button indicates that more information is available. When a user pushes the blue button they get step-by-step instructions on how to perform CPR.

The overall impact of these products is too recent to quantify, but simulations and training exercises have been conducted that do measure the effects. For example, in May 2006, 400 first responders and volunteers from 28 public safety agencies participated in a Homeland Security training simulation in which terrorists occupied an airport hangar at the Long Beach Airport.

As the simulation progressed, one of the terrorists escaped from the hangar, hijacked a fuel tanker, and intentionally crashed into a jetliner full of passengers. The simulated collision injured 100 passengers on board and started a raging fire threatening the other passengers.

The first responders used as new handheld performance support product based on mesh networks—wireless networks that can be created on the fly—developed by a company called PacketHop. The technology tracks the position and movement of peers and suspects, provides navigational assistance, provides a white boarding feature allowing multiple people to collaborate on strategy, includes an instant messaging interface, and can access video from any wireless-enabled video camera in the area. The performance support is both automated content and peer-to-peer coaching and collaboration.

PacketHop reports that the product enabled the first responder teams to rapidly coordinate a unified response and they were able to extinguish the flames before anymore people were injured. It allowed them to quickly coordinate the rescue of the passengers and provide emergency aid to the injured passengers.

Bottom line: Precision performance support requires high-fidelity behavior models. Developing these high-fidelity behavior models takes a great deal of work, but in many situations it is worth the effort.