Go back to bed; overnight intensivists don't reduce mortality - PulmCCM
Dec 252012

We intensivists are a scarce and valuable commodity -- just ask us, we'll tell you. It's been consistently shown, for example, that involved intensivists in an ICU during the day correlates with improved mortality and efficiencies in care. Now that's something we can all cheer (ourselves) about.

Many intensivists also like to sleep, and may have found it a bit harder to do so as certain academic leading lights, voicing support of the business-supported Leapfrog Group, have called for 24/7 intensivist coverage (overnight, in-house) to become the standard of care. The evidence for benefits of this are much less robust, mainly limited to 2 single-center "before/after" studies:

  • Gajic et al, at Mayo (Crit Care Med 2008), suggested 24/7 intensivists improved adherence with evidence-based guidelines and reduced ICU complications, possibly reducing length of stay by a day.
  • Blunt et al (Lancet 2000) concluded that 24/7 intensivist coverage may have reduced mortality at Queen Elizabeth Hospital, Norfolk UK (using the soft-ish standardized mortality ratio).

Meanwhile, others have pointed out how much it would cost to implement 24/7 intensivist coverage, and also how bad it would suck (e.g., increasing burnout rates).

So thank goodness for David Wallace, Derek Angus, Amber Barnato, Andrew Kramer, and the great Jeremy Kahn for coming to the rescue with a study in the May 31 New England Journal of Medicine that can help us all sleep a little better ... at home.

What They Did

Authors performed 2 retrospective cohort studies. Unlike the before-and-after / historically controlled single-center studies mentioned above (which are subject to bias from changes through time or the Hawthorne effect); this study compared ICUs with and without nighttime intensivist staffing at roughly the same point in time.

The main cohort included 25 hospitals (49 ICUs, 65,752 ICU admissions) that participate in Cerner's APACHE data collection program, cross-referenced against a survey of those same hospitals that included responses on ICU physician staffing:

  • 12 ICUs had 24/7 nighttime intensivist staffing (22%).
  • 37 ICUs did not have 24/7 in-house intensivists (78%). Most of these (25) had residents in-house overnight; the rest had either non-intensivist physician coverage overnight (5), a midlevel (1), or no in-house overnight coverage at all (6).
  • 22 of these ICUs (45%) had "low-intensity" daytime staffing (intensivist consultation was optional in the ICU) while 27 had "high-intensity" daytime staffing (mandatory intensivist consultation, or intensivists as the primary team).

There were no significant differences detectable on paper among the patients admitted to "overnight" vs. "non-overnight" staffed ICUs (APACHE score, need for mechanical ventilation, demographics, etc).

The other "verification cohort" included 107,000 patients admitted to ICUs in Pennsylvania hospitals only, and used the same methodology.

They report their confidence intervals account for statistical clustering at the ICU level (a concern when comparing groups, rather than patients).

What They Found

Unadjusted in-hospital mortality was ever-so-slightly lower in ICUs with nighttime intensivists (12.8%) compared to those without (13.4%), with a barely non-significant p=0.053. At first glance, that's an absolute difference of 0.6% -- meaning for every 166 people admitted to ICUs staffed overnight by intensivists, a life might be saved (or at least extended throughout a hospitalization).

However, all the observed potential benefit was limited to ICUs that did not have "high-intensity" daytime intensivist staffing (mandatory intensivist consultation, or intensivists as the primary team):

  • In those 22 ICUs with low-intensity daytime staffing, having an in-house intensivist overnight resulted in an odds ratio of 0.63 for in-hospital mortality.
  • In the 27 ICUs with high-intensity daytime staffing, nighttime intensivists had no apparent effect (odds ratio for mortality 1.08).

The findings were duplicated in the "verification cohort" using the Pennsylvania hospital data: nighttime intensivist coverage was associated with reduced mortality only in ICUs that did not use "high-intensity" intensivist staffing during the day.

What It Means

Nighttime intensivist coverage was not associated with a reduction in in-hospital mortality in ICUs that required intensivists be consultants or primary attendings for all critically ill patients during daytime hours. In ICUs that did not require this "high-intensity" daytime intensivist coverage (already known to be beneficial), nighttime intensivists did apparently save lives.

Although the economic crisis has probably taken 24/7 intensivist coverage off the table as a priority for many medical centers for now, simply due to the increased cost, with this study the burden of proof is on the night owls to show that adding 24/7 intensivist coverage to adequately daytime-staffed ICUs really does provide benefits that are worth the costs -- economically and with regard to "sustainability" within the profession.

No one has pointed out what I could tell you given 48 hours, a list of medicine residents' email addresses and SurveyMonkey: making 24/7 in-house intensivist coverage a standard would result in a sudden and precipitous fall in the number of quality physicians choosing critical care as a career. That would be no good for the profession, or for our patients.

Update: A study in NEJM 2013 showed no improvement in outcomes during periods of overnight intensivist staffing at the University of Pennsylvania over the course of one year, as compared to the weeks those attendings were available by phone, with residents in the ICU.

Clinical Takeaway: For administrators, consider making intensivist consultation mandatory on all patients admitted to the ICU. For intensivists ... sweet dreams!

Wallace DJ et al. Nighttime Intensivist Staffing and Mortality among Critically Ill Patients. NEJM 2012;366:2093-2101.

Liked this post? Get a weekly email update, and explore our library of clinical guidelines, practice updatesreview articles. and board review questions.

PulmCCM is an independent publication not affiliated with or endorsed by any other organization, society or journal referenced on the website. (Terms of Use)

Authors: contribute your work in a guest post.

  5 Responses to “Go back to bed; overnight intensivists don’t reduce mortality”

  1. So, if we don’t require ICU physicians at night for patient care, maybe the ICU physician will become a passing fancy. Just remember the years of training and studying that we have all endured, to become the last hope for those critically-ill patients, needing a very different professional evaluation and treatment. The efforts of ICU physicians and nurses are the best of the best for acute health care delivery. We should all want a knowledgeable ICU physician at our bedside, either in-person or eICU, to deliver the best possible care. Our goals must continue to be critical thinking and compassionate care! Please don’t marginalize or minimize our professional efforts or the value of our skills. And yes, we DO need sleep at night.

  2. Dr Weinberg, thank you for these insightful comments.

    I agree with all your points. I would never marginalize our efforts or our skills’ value — in fact, I value my own quite highly 🙂

    I would go even farther, in principle: the ideal care model would be an individual ICU physician sitting in each patient’s room, or maybe just outside, during the entire ICU stay. It’s not that absurd to suggest; this is more or less what ICU nurses do (with 2 patients).

    But we’ve already accepted that this is an infeasible model, mainly for economic and resource considerations: there simply aren’t enough docs, and we cost too much. Once one has accepted those constraints, considering other real-world factors like lifestyle and sustainability of the profession is perfectly legitimate and should not be given short shrift.

    The good news is, the best data we have shows that in properly staffed ICUs where well-trained intensivists develop thoughtful plans and communicate them well during the day with the supporting team (who actually deliver most care), the intensivist’s physical absence overnight does not mean worse outcomes. Would I want an intensivist at the bedside at 4 AM for myself or my family member, God forbid one was needed? Of course! I would also like to have that physician’s personal cell phone number that I may call at any time, and for her to never leave the hospital, or to ever hand off care to another physician. But these are not realistic requests, nor are they necessary–and one could argue, neither is the physical presence of the intensivist overnight.

    eICUs are an interesting alternative and I look forward to seeing how they play out (but not to working in one).

    ICU physicians will never go out of style or become a passing fancy — if anything, I expect we will come into higher and higher demand as the lifestyle of the job becomes less desirable and fewer top residents subsequently choose critical care. Agitating for 24/7 in-house coverage is a surefire way to drive down the quality and long-term health of our profession, over the long term.

  3. I have to say our intensivist program has greatly improved with 24/7 coverage but this has only been able to happen with NP intensivist program added to Nights. It has improve quality of patient care and shorter stays in the ICU. The physician is available for us to talk if needed but at home and comes in only in a emergency.

  4. Congratulations on another great review. In Argentina we face similar staffing problems, needless to say very little proportion of our ICUs gets an intensivist 24/7. Another turn for good care and good sleep is to implement 12 hours shift, alongside with intensive daycare intensivisr coverage. Will share this post on T. Thanks and keep up the good work.

  5. It’s worth mentioning that a very common community ICU model (daytime in-house intensivist coverage, no one in the ICU overnight) is greatly underrepresented in this study.