Stroke Data Hint at High Practice Variability Between EMS Agencies

Individual EMS agencies have varied substantially in their stroke case documentation, raising the question of whether problems might also extend to their operations and implementation of stroke protocols, according to preliminary work from Michigan.

Based on 147 EMS agencies working in the state, their compliance with documenting time since last known well (LKW) for stroke patients was a mere 24%. They performed somewhat better at approximately 50% compliance for documenting prehospital stroke scale (PSS) scores, stroke recognition, on-scene time ≤15 minutes, and hospital prenotification. Documentation of a glucose check reached a high of 82%.

“We found that variation in each measure could be attributed in part to the transporting EMS agency. For measures such as on-scene time and EMS stroke recognition, this contribution was relatively small. On the other hand, >50% of variation in PSS and LKW documentation was attributable to the transporting agency,” reported J. Adam Oostema, MD, MS, of Michigan State University in Grand Rapids, and colleagues.

While this may reflect true practice variation between agencies, the problem may also stem from differences in EMR software that affect data collection by the statewide EMS registry, they noted in the Journal of the American Heart Association.

“This is a fascinating and concerning issue, with major implications for the EMS community,” according to an accompanying editorial by Rebecca Cash, PhD, of Massachusetts General Hospital and Harvard Medical School in Boston, and Christopher Richards, MD, MS, of University of Cincinnati College of Medicine.

They suggested that EMS clinicians may already be documenting much of the necessary information for stroke performance measures in a free-text narrative report, which then has significant information lost when carried over to large research and quality improvement databases.

“Of course, compliance with core metrics may reflect gaps in knowledge and compliance with protocols; however, with the data we have available, knowing whether these results reflect data capture limitations or true deficits in clinical care is challenging,” according to Cash and Richards.

“Harnessing the power of machine learning, artificial intelligence, and other evolving technologies could bridge the gap between clinical documentation in EMS electronic patient care records and discrete data fields in large databases, opening the potential to better measure and track care currently provided in the prehospital setting. Only when accurate data are known can true clinical performance be assessed and improved,” they concluded.

Oostema’s group nevertheless suggested that there are opportunities for improvement in EMS stroke screening, recognition, and prenotification in time-sensitive stroke care.

For their retrospective study, they linked data from Michigan’s EMS Information Systems and the Michigan Acute Stroke Registry. The dataset covered 5,707 confirmed stroke patients admitted to one of 38 hospitals after transport by one of 147 EMS agencies. Cases included ischemic stroke, hemorrhagic stroke, or transient ischemic attack from Jan. 1, 2018 to June 30, 2019.

Included stroke patients were largely over age 60, with 52% being women. About one in three patients had onset-to-door times within 120 minutes, and 18.5% arrived within 60 minutes.

Older age, higher National Institutes of Health Stroke Scale, ischemic stroke, and earlier presentation were associated with more compliant prehospital care for almost every measure.

“We suspect that more ‘obvious’ stroke cases (severe strokes and those with symptoms such as unilateral weakness) are more likely to be correctly identified and treated appropriately by EMS, as previous studies demonstrate,” the investigators said. “Overall, these results are encouraging that patients most likely to receive optimal prehospital care are also the patients most likely to be candidates for acute ischemic stroke treatments.”

Oostema and colleagues noted that their study may not representative of all prehospital stroke care in Michigan, however. The dataset covered 38 of the 104 acute care hospitals in the state and 147 of the 281 transporting EMS agencies; participating hospitals tend to be larger, certified primary or comprehensive stroke centers in urban or suburban areas.