HIMSS14: Incorporating human factor engineering in health IT

ORLANDO—Human factor engineering can significantly impact the usability of an EHR, according to a panel during the Patient Safety Symposium of the Health Information and Management Systems Society annual conference.

“Often with implementation we work as imagined--not as performed,” said Terry Fairbanks, MD, MS, director of the National Center for Human Factors Engineering in Healthcare (NCHFEH) at the MedStar Institute for Innovation. Rather than fitting the IT to the workflow, “the idea is we can educate users so it will work.”

White boards are common in almost every emergency department and operating room in the western world, he said. They are a “cognitive artifact of work developed over time by front-line workers.” They continue to be used more than IT systems “not because people don’t want to use computers but because the white board has the functionality needed.”

He cited an 18-month qualitative study on white boards to see the data displayed and used and by whom. “It’s very rich. There is so much information there.” However, very little of that information is translated into the health IT system. For example, a simple circle indicated to techs that a room was ready to be stocked. Everyone began to notice that rooms weren’t being stocked but nobody knew why.

“If we want to move to the next level, we have to understand that to design for this kind of work takes a lot of resources and thousands of permutations.”

Raj Ratwani, PhD, scientific director at National Center for Human Factors in Healthcare, shadowed emergency medicine physicians to track efficiency before, during and after EHR implementation. Time with patients and paper both go down at go live and post-go live, time with patients go back to where it was before but so does time with paper. With EHRs, more users are doing 3 or 4 tasks within the same minute. That trend, he said, “is very troubling.”

IT support is very important, he said, particularly IT staff that understands EHR complexities. Communication between the vendor and IT staff as well as between front-line users and IT staff is essential, he added.

Hardware matters, Ratwani said, including screen size, process and network speed. “It’s easy to lose trust in systems if there is a slow response.” A slow response also encourages multitasking which “introduces memory issues that may result in error and certainly increases the probability of error.”

Organizations must evaluate how people are using the system, Ratwani said. “Look for work-arounds as an indicator of a need not being met. Eliminate options that are not used. Look at cancelled orders. These are good indicators of where the EHR is not living up to its promise.”

Many organizations skimp on training, he said, because it comes at an added cost. However, “There is a clear relationship between investment in training and physician performance.” Training needs to happen on a system the same as the one being implemented and it can’t happen too early so trainees forget what they’ve learned by the time of the go live. Training also should involve realistic training scenarios covering the complete workflow as well as phased training to introduce more advanced functionality. Organizations that don’t incorporate these guidelines “will pay for it later in unfortunate ways,” said Ratwani.

To avoid problems with ordering, terminology should match the ways people discuss tests, said Zach Hettinger, MD, MS, director of informatics research for MedStar's National Center for Human Factors Engineering in Healthcare. Many systems work on a many-to-one strategy, and default to common frontline terms. Systems should work with role-specific language, Hettinger said, because nurses need to know how to deliver and pharmacists need to know how to mix. “Asking staff to communicate in another lexicon creates potential for error.”

He provided an example of how frustrating ordering can be. Inputting urine sodium yields no results. Neither does “urine NA” and “NA urine.” Finally, “NA level urine” appears. “We’re training the users to use the system instead of the system to support users.”

Order sets are living documents, Hettinger said. Organizations should obtain ongoing feedback from frontline workers, review orders that are commonly added to order sets, review orders that are not ordered through sets and look for the reasons why, and review canceled orders looking for trends and near misses.