Hospitalists Are Key to Rapid Response to In-Hospital Stroke

For the past 6 years, Dr. Ethan Cumbler has been immersed in figuring out how to reduce the time from when the medical staff first notices signs of new neurologic deficits in a hospitalized patient to when the patient receives an evaluation. Before he and his colleagues at the University of Colorado Hospital in Denver first started the project in October 2006, it took 4 1/2 hours, on average. Today, the average time to evaluation for an in-hospital stroke is 30 minutes, similar to the response time in the emergency department.

Dr. Cumbler credits their success, in part, to the formation of an in-hospital stroke alert program, in which the medical staff calls in a multidisciplinary stroke response team rather than the patient’s primary medical team. Earlier this year, the Society for Hospital Medicine awardedDr. Cumbler, and the multidisciplinary team at the University of Colorado School of Medicine, the Award for Excellence in Teamwork in Quality Improvement for their work on improving in-hospital stroke response and studying it.

In an interview, Dr. Cumbler, who is also the director of the University of Colorado Hospital Acute Care for the Elderly service, explained the essential elements to reducing response time and the critical role for hospitalists.

Question: How are most hospitals doing in terms of in-hospital stroke response time?

Dr. Cumbler: Hospitals are all over the map in terms of their sophistication in dealing with inpatient strokes. When I speak at hospitals around the country, I generally hear about one of three different types of responses to a patient with new neurologic deficits on the floor.

The first one is the traditional response, in which there is not a systematized approach to a patient with new neurologic deficits but rather the call goes to the primary team. You have to recognize the huge variability in stroke experience, knowledge, and comfort when you’re talking about calling any physician who might have a patient anywhere within the hospital.

The second response that I hear about is a rapid response team that is able to initiate a stroke evaluation. That’s not the path that we use here at the University of Colorado Hospital, but for many hospitals, this works out extremely well and ties in nicely to their existing system of rapid response for cardiopulmonary crisis. This does however create the imperative to make sure the rapid response team members are stroke trained.

The last approach is the system that we use here at the University of Colorado, which is a dedicated inpatient stroke alert system. For hospitals that use this approach, the response team is usually the same as the one that responds to strokes in the emergency department.

Question: Is there room for improvement?

Dr. Cumbler: There are two places where there is room for improvement. The first is response time. This is a classic opportunity for the application of Lean quality improvement methodology to reduce complexity, to eliminate steps which don’t add value, and to reduce variation. When we look at the data both from the National Stroke Association’s QI initiative, as well as from a Michigan stroke registry, we find that very few patients meet the goal of 25 minutes from recognition of stroke symptoms by medical staff to CT scan, if the stroke happened in the inpatient setting.

The second place where there’s room for improvement is in adherence to quality metrics. I’m actually doing research right now with the American Heart Association using the Get With the Guidelines national database to examine quality of care and outcomes for in-hospital stroke compared to those in the community. This is the largest analysis that’s been done on in-hospital stroke by a factor of about 30

Question: Are hospitalists equipped to respond to strokes?

Dr. Cumbler: I would argue that the hospitalist is equipped, or could be equipped, to be part of that response team for in-hospital strokes. In part because I think stroke is within the circle of expertise of hospitalists. But also because we have to recognize that not all stroke alerts are for stroke. When we have looked at this across a number of different analyses, ours and others, we find that somewhere between about 30% and 50% of all stroke alerts, are not for stroke. They are for one of the stroke mimics, such as severe hypoglycemia, or narcotic overdose, or delirium from a new systemic infection. Who is better to assess that new acute neurologic change in a complex, ill inpatient than the hospitalist?