Sunday, April 22, 2012

Do ICUs make a difference?

Here is an interesting article from the Archives of Internal Medicine, sent to me by Marco D. Huesch, MBBS, PhD, of the Leonard D. Schaeffer Center for Health Policy and Economics at the University of Southern California.  The article is by Henry Stelfox and others and is entitled "Intensive Care Unit Bed Availability and Outcomes for Hospitalized Patients With Sudden Clinical Deterioration."

Looking at a cohort of 3494 patients in Alberta, the study team investigated whether clinical outcomes varied depending on whether such patients were treated in the ICU versus the regular wards. In a nutshell, there was no difference in in-hospital mortality for these inpatients with sudden clinical deterioration, when they were effectively randomized (as a result of temporal variation in ICU capacity) to receive ICU referral or ward treatment. 

Perhaps this suggests that, in this study, many patients who are too well to benefit -- or too sick to benefit -- from ICU care tend to occupy ICU beds. Does the data about the latter group have implications for our efforts to improve critical care?  Does the data about the former group suggest that we are incurring the expense of ICU treatment when it is not always necessary?

I'd love to hear from you critical care folks out there.

2 comments:

ICU doctor said...

At the risk of sounding defensive (I am an ICU doctor), I think it is worth noting that the primary hypothesis being tested here was NOT if ICUs have a benefit. It was if not having an ICU bed available was bad for patients - an exploration of triage and rationing and its effect on patient outcomes. So, at best, this conclusion about the value of ICUs is a secondary one - worthy of generating additional hypotheses. The increase in the percentage of patients having their goals of care changed to comfort when no ICU beds are available and suggests how ICUs may actually function - as a mechanism of system resilience.

Suppose I am a busy hospitalist providing care to a multitude of sick, but not ICU patients. I am occupied with addressing my patients' acute clinical deterioration, adjusting their medications, coordinating their care among multiple specialties and so on and so on. Ideally, I would also be discussing overall goals of care if the patient does not respond to therapy. However, this is a hypothetical and I have patients with real problems who more than fill my time. This is analogous to the difference discussed in behavioral economics works related to charitable giving - the difference between identifiable and statistical victims (see J Risk and Uncertainty 1997; 14: 235-57). We respond to the former - not the latter. However, when a patient deteriorates, we now have an identifiable victim and have to act. What is the action that allows me to get back to caring for the other 20-30 patients I have in the hospital? If there are ICU beds available, perhaps I send them to the ICU for care - or discussions of goals. If there are no ICU beds, I can either continue to try and manage them on the ward, with fewer resources and other patient needs piling up - or I can discuss the goals of care now. This may sound like we are making decisions based on what is easiest for us to do - but in times of uncertain action, this is a compelling incentive for ALL human behavior.

Now, what would be very interesting is to examine the mortality of the hospitalist's other patients. Are they more likely to deteriorate as a result of the time he/she spends away from their care at the bedside of a critically ill patient who needs an ICU but can't have one yet? A similar effect may happen to the other patients whose primary ward nurse is now committed to the bedside of a single patient - as their care is hurriedly handed off to another nurse or their care is delayed. Other outcomes which might be affected more than the mortality of the deteriorating patients are the satisfaction and burnout of the hospitalist and the ward nurse.

I think this is a very interesting paper. For me, it illustrates yet again that there is a major impact of decisions made by imperfect decision-makers (physicians) that affect patient outcomes. The factors that influence these decisions are under-studied relative to molecular medicine - but, perhaps, more influential and more easily influenced.

Or perhaps I am just being defensive....

Jack said...

What about the externalities of providing ICU-level care in a step-down type nursing unit?

One of the main benefits of ICU admission is 1:1 patient care and dedicated support personnel (RCPs, Pharmacists, Radiology, etc). Most medical equipment is mobile within the medical facility and thus can be brought to any bedside.

When a high-acuity patient receives care in a low-acuity setting, he will inevitably consume more resources than he would in the ICU. A step-down unit with a 4:1 care ratio will have to be altered in order to provide 1:1 care for the deteriorating patient. Thus, staffing levels must be adjusted accordingly, adding workload to on-duty staff members or calling-in additional staff.

While there might not be a change in mortality (which speaks highly of the facility's adaptability), it is probably much more costly, in terms of money and resources, to provide ICU-level care in a non-ICU unit.