Antibiotics: There's no free lunch

A new, important paper in JAMA Internal Medicine from Sara Cosgrove's group at Johns Hopkins demonstrates the collateral damage of antibiotics. In this retrospective cohort study of 5,579 internal medicine inpatients, 1,488 (27%) received a parenteral or oral antibiotic for at least 24 hours. The most common indication for antibiotics was UTI, Adverse events due to antibiotics were captured over the 30-day period after antibiotic initiation, with the exception of C. difficile infection and MDRO infections, which were captured over the ensuing 90 days. Median duration of therapy was 7 days. Of the patients treated with antibiotics, 19% had no clinical indication for antibiotic therapy, and 20% developed at least one associated adverse event. The breakdown of adverse events is shown in the visual abstract below (note: for this analysis, I combined the 30- and 90-day outcomes). We are becoming more cognizant that antibiotics are not benign therapies. Kudos to Sara and her colleagues for their work in raising our awareness.

Guest Post: Thinking about Contact Precautions

Anthony Harris, MD MPH

This is a guest post from Dr. Anthony Harris, Professor of Epidemiology and Public Health, University of Maryland School of Medicine.

I have a lot of admiration for this blog and in the spirit of academics, I would like to share my somewhat different interpretation of the contact precautions literature. I offer three points for your consideration:
 
1. Contact precautions do not lead to an increase in adverse events: 

Instead of the frequently cited small observational studies, I think it is most important to focus on the one randomized trial that evaluated adverse events associated with contact precautions (I acknowledge my bias in that it is the study I led). A randomized trial should most often trump observational studies especially since it's near impossible to control for confounding by indication, i.e. why was the patient placed on contact precautions. In fact, the latest analysis of our randomized trial data showed a trend towards decreased adverse events in the universal contact precaution arm. (See: Croft L et al. Clin Infect Dis. 2015 Aug 15;61(4):545-53). To quote from the Conclusion: "Concerns of adverse events resulting from universal glove and gown use were not supported." So we should be clear that other than the cost issues of gloves and gowns no high level study has shown any adverse events from contact precautions. Healthcare workers do go into the patient room less often when the patient is on contact precautions but this has not been shown to lead to an increase in adverse events. In my experience (and the data supports this), healthcare workers just bundle their activities and thus perform the same activities in the room with fewer visits.

2. Methodological problems in studies that have removed contact precautions: 

Studies that show no effect of removing contact precautions have serious methodological problems. The largest problem is that they are incredibly under-powered. The studies that show “no difference when you remove contact precautions” are too small to detect a difference and thus may falsely conclude that removing contact precautions is safe.

3. Need a better solution before removing the current standard of contact precautions: 

As much as I would love to stop wearing gloves and gowns, antibiotic-resistant bacteria are a continuing problem. They are not going away and other than MRSA, they are not decreasing. Until we have better solutions, I believe that we should not be abandoning contact precautions. This is particularly true in high risk settings such as the ICU. In the ICU where acquisition of an MDRO leads to infection 20-30% of the time during the index ICU admission, the stakes are too high not to prevent patient-to-patient transmission. I believe that the phase 0, phase 1 and phase 2 data on contact precautions are strong and the biologic plausibility that they prevent patient-to-patient transmission so strong that I don’t think we should abandon contact precautions based on underpowered "removal studies" and an adverse-event literature with few studies with strong internal validity. Of course there are certain settings and certain bacteria where removing contact precautions might make sense and these scenarios should be studied using large, sufficiently powered and methodologically sound trials.

"I am isolated here..."

The October issue of Infection Control and Hospital Epidemiology has an interesting paper on contact precautions by the University of Maryland group. This prospective cohort study of over 500 patients used standardized interviews to compare perceptions of inpatients who were cared for under contact precautions versus those who were not.

Patients who experienced contact precautions were twice as likely to report problems with their care. Specifically these patients noted poor care coordination and a lack of respect for their needs and preferences.

One particular quote from a patient stands out:

“I am isolated here. When people put on the gowns, I feel dirty and alone… [They] even had to wear them when I was being wheeled around for tests.”

So in a nutshell, another study tells us that contact precautions is a patient unfriendly intervention.

Photo:  Bob Tymcyszyn/QMI Agency in the Toronto Sun.

Do Contact Precautions Cause Depression?

STOP Contact Precautions?

Mike has written several time about his concerns for the side-effects of contact precautions. (see his kill contact precautions, personal view and adverse effects posts). His post "adverse effects" discusses Dan Morgan's systematic review that looked at the state of the literature measuring what has become dogma for many hospital epidemiologists and clinicians: contact precautions harm people. After completing this review, we weren't entirely comfortable with the literature, so we set out to complete our own series of studies seeking to assess the association between contact precautions and adverse outcomes. 

The first such study by Hannah Day, a PhD student working with Dan Morgan and me, has just been published in the March 2011 AJIC.  It was a pilot study that allowed us to gather baseline estimates to complete power calculations for the larger studies that will follow. She measured the baseline levels of depression and anxiety in patients admitted to non-ICU wards of an acute-care VA hospital. Of note: this study was a sub-study within an MRSA prediction-rule study that we'd published earlier and Dan Morgan posted on a few months ago. (see Veteran's Day MRSA post)

In 2009, 103 patients (20 on contact precautions and 83 unisolated) were approached within 48 hours of admission and received a questionnaire that included a body of questions called the HADS - Hospital Anxiety and Depression Scale. What did she find?  The HADS score was 2.2 points higher in patients on contact precautions (p=0.21).  The odds ratio for having either depression or anxiety was nearly twice as high on contact precautions (OR=1.87, 95% CI 0.61-5.69).  Yes, the study was under powered.  Intriguingly, the increased HADS scores associated with contact precautions exposure was higher in those patients newly isolated (<1 year since first isolated).

Before everyone gets too excited, what does a HADS score difference of 2.2 mean? Well, this study was too small to determine whether this increase was due to depression or anxiety.  Additionally, with the combined scale, a minimum difference of 3.0 is considered clinically significant.  Thus, even if there is an association, it may have little clinical meaning.  Finally, the HADS was measured on admission, so we were unable to determine if isolation "caused" the depression. To do that, we would need to do repeated measurements on admission and throughout the stay to see if there was a change in HADS later in the admission.  More exciting data to come...

Day HR et al. Am J Infect Control March 2011

DHHS releases 2010 Adverse Events in Hospitals report

The Department of Health and Human services just released an 81-page report titled: Adverse Events in Hospitals: National Incidence among Medicare Beneficiaries. Using a nationally representative random sample of 780 Medicare beneficiaries discharged in October 2008, physicians determined (1) whether an adverse event occurred, (2) whether the event was an NQF Serious Reportable Event or a Medicare hospital-acquired conditions, (3) what the level of harm was to the patient, and (4) whether the event was preventable.  Using this sample, they estimated that 13.5% of all hospitalized Medicare patients experienced an adverse event and in 1.5% the AE contributed to their deaths.  These extrapolate to 134,000 adverse events and 15,000 deaths in a single month. Multiply by 11.7498 (or 365.2425/31) if you want yearly estimates for an average year.  Yes, I'm being a smart a**, but multiplying by 12 is incorrect.

One thing we're always concerned about on this blog is the percent of HAIs that are actually preventable in the current 'get to zero' world that we live in.  In the DHHS report, physicians estimated that 44% of the AEs were preventable, 51% were not preventable and in 5% they were unsure.  The costs were $324 million in October 2008 or 3.5% of all hospitalization costs. They suggest that the FY2009 attributable costs of AEs were $4.4 billion, with two-thirds of the costs being associated with extended hospital stays.

So what about HAIs?  Table 3 in the report classifies the 128 AEs into categories and 19 or 15% of the AEs were HAIs.  Thus, 2.5% of all hospitalized Medicare patients had an HAI. There were 5 UTIs, 4 CLABSI, 4 other BSI, 4 RTIs and 2 SSIs.  The physicians classified only 60% of the infections as preventable.  I wonder if this will help Mike achieve one of his wishes for the 2010 New Year? I could probably dig deeper but I've got work to do and I don't want to ruin all of your fun.

Link: November 2010 DHHS Adverse Events in Hospitals report.

h/t: Megan McKenna