James S. Todd Memorial Research Award

Interruptions in the Emergency Room

Nancy Franklin, PhD, Stony Brook University

ER physicians typically juggle several tasks at once and are required to wait for test results, colleagues, and other resources whose availability is unpredictable. Traditional research in both medicine and cognitive psychology suggests that performance in such an environment should benefit from a total elimination of interruption. We are examining quite a different view, by focusing on the potentially critical role of interruptions that serve as optimally-timed reminders to return to previously suspended tasks.

Our ongoing project consists of three studies. (I) Observations of ER physicians on shift have shown that interruptions are abundant and are constituted largely of relevant interruptions, which (while they may disrupt the local task) may lead to more efficient global functioning. We have also found differing patterns of interruptions for physicians of different levels of experience, reflecting their different roles in the ER. (II & III) Two demanding video game-style simulation tasks ask undergraduate participants to solve various criminal cases or ask ER physicians to follow medical cases requiring urgent care. These simulations allow us to both experimentally manipulate factors relevant to interruption and record participants’ actions in great detail. The data collected thus far suggest that interruptions that call people back at the appropriate time to a previously suspended task may impact performance profoundly, and in the direction opposite to what the traditional research literatures (which have focused on irrelevant interruption) have found. Our work may help to guide policy and training for personnel so that they can share information and prioritize tasks even more efficiently. The support from NPSF has allowed us to develop a complex, flexible analogue to cognitively taxing work environments. This tool will provide much greater experimental sensitivity for investigating a problem that impacts not just the ER, but also other demanding healthcare and professional settings.

NPSF Board Grant

Knowledge Discovery: The development of an error/solution matrix to improve patient safety

Harold S. Kaplan, MD, Columbia University College of Physicians and Surgeons

The overarching goal of this research project was to turn medical event data into information and actionable solution knowledge. We hypothesized that information embedded in medical event report narratives represents an untapped source of solutions to human and system failures, and that analysis of these narratives can contribute to the development of a new taxonomy of solutions. An error/solution matrix could then be developed that would aid in formulating solutions based on event characteristics.

Recovery, mitigation, and corrective action data from a set of near-miss and no-harm medical event reports were analyzed. The majority of ‘solutions’ found in the narratives were either expected (e.g.: be more careful, follow the protocol), or specific to a certain event (with its associated contextual nuances), and would have resulted in solution categories neither novel nor ultimately useful. Solution taxonomies from fields outside of healthcare were then studied, but no single existing typology from industry was found to be optimal for classifying potential solutions for use with medical events. We needed to create a new taxonomy that would combine what we had learned from both medical event reports and other industries.

Three sub-taxonomies were developed: (1) problem descriptors (e.g.: incorrect, confusing, damaged), (2) the condition involved (e.g.: action, item, policy), and (3) solutions and their subsets (e.g.: make obvious, simplify, automate). These subtaxonomies were then linked in matrices and tested against medical event data.

We are continuing the evaluation and evolution of the matrix, and adding solutions as we mine the data. Although very useful as it currently exists, preliminary data suggests that the matrix may be simplified or further revised. We are also testing a preliminary method to computerize the matrices for ease of use. The utility and potential impact of this work will be determined by how effectively it strengthens the weakest link in the healthcare problem-solving paradigm – the generation of optimal solutions.