sinaiem.org

Jack tackled a controversial topic at this month’s Journal Club — what’s the evidence for giving antibiotics within four hours for CAP patients? It’s a good question, because how well we perform at this task is a big part of how our hospitals are measured. Ineed, pneumonia antibiotic timing is one of JCAHO’s Core Measures and there are only more such metrics down the road — so we’d like to think that our funding depends on rock-solid science and proven benefits.

Well…

As Jack noted, the 4-hr policy is based primarily on four papers,

1) Kahn et al, JAMA 1990 Oct 17 264(15) 1969-73 with comments 1995-6

2) McGarvey et al, Quality Review Bulletin April 13(4) 124-30

3) Meehan, Houck et al.  JAMA 1997 Dec 17 278(23) 2080-4 

4) Houck et al, Arch Intern Med. 2004; 164(6):637-644.

It’s this last paper we’re going to discuss — a retrospective study derived from a national sample of medicare patients with pneumonia.  

Full disclosure — the lead author, Peter Houck, works for the Oklahmona Foundation for Medical Quality. Sure, they’re not-for-profit, but there’s an inherent conflict-of-interest. It’s kind of like how JCAHO reviews hospital performance, but also has a consulting arm that avises hospitals on how to pass their review, for a fee. The line between “the patient’s best interest” and “the regulatory agency’s best interest” becomes blurry, when someone from a quality-assurance agency is researching measurable hospital policies.

Since the ‘therapy’ being tested here is timing to administration of antibiotics, this study can’t ethically be done prospectively. Their retrospective sample is worth noting, however: They took their numbers from a database of Medicare (over 65) with a primary or secondary ICD-9 code of pneumonia sampled from July 1- Dec 31, 1998, or Sept 1, 1998-March 31, 1999 (this database had 346,105 patients, though it seems possible they sampled the same patients twice). The authors randomly selected up to 850 cases from each state (more like the Senate than the House of Representatives — Wyoming’s handling of pneumonia could be weighed equally with California’s), giving a study set of 39,242 patients .

Exclusion criteria included death or discharge on day of admission, age under 65, lack of radiographic evidence of pneumonia, history of immunocompromised status, recent hospitalization, or lack of ABx timing documentation. This eliminated over half their set, leaving 18,209 patients from 3732 hospitals (avg = 3/hospital, but one hospital accounted for 122 patients in the sample).

Their outcomes were in-house mortality, mortality within 30 days of admission, LOS in house, and readmission within 30 days of discharge. Jack recommends we start assessing the data with Table 4, which reports 30-day mortality, divided up by who got ABx when. So patients that got antibiotics within one hour, compared to those that got ABx after one hour, showed no significant difference in mortality. Ditto with two hours. But those patients that got ABx within 3 hours had a mortality of 11.7% (95%CI 11.0-12.6) compared with 12.3% (95%CI 11.5-13.0) for patients getting antibiotics after 3 hours, giving an adjusted odds ratio of 0.88, which was significantly different. Three hours! So why isn’t that the rule? Because, ah, four hour cutoffs are “commonly used in quality improvement activities.” OK…

 The benefit of ABx before X hours, compared to after X hours, persists for values of X from 3 to 9 (if you don’t get antibiotics before 10 hours, it makes no significant difference in mortality if you get them after 10 hours.

Since the authors decided to focus on the 4 hour cutoff, they generated a lot of data around that timepoint. Their patient characteristics (Table 1) shows sufficient randomization in many categories, though it looks like women, minorities, and those over 85 were more likely to get ABx after 4 hours, rather than before. In Table 2, we see that city hospitals were less likely to get ABx administered before four hours, compared to nonmetropolitan hospitals. The same slowness was seen true in for-profit hospitals, as well as teaching hospitals. Make of that what you will.

Table 5 shows the breakdown of all outcomes, depending on whether ABx were given before or after 4 hrs. Length of stay, in-house mortality, and 30 day mortality are all significantly lower in the “within 4 hr” group (30-day readmission is unaffected). Looking closer at 30-day mortality, we see that 12.7% of patients die with antibiotics after 4hr, compared with 11.6% who die with antibiotics before (confidence intervals do not intersect). That 1.1% difference means the NNT is 91  (if we give 91 patients ABx before 4 hrs, we’ll save one more life). Of the 600,000 medicare patients admitted for pneumonia each year, that translates to 6600 lives saved.

Also striking about Table 5– there’s a huge mortality difference between PSI (pneumonia severity index) class IV and V, compared to class II and III… but all patients who got ABx within 4 hours fared better than those who did not, in terms of in-hospital mortality, 30-day mortality, and shorter LOS.

So, early administration is obviously a good thing, right? Well, no – it turns out that patients who got antibiotics before two hours actually did worse (see Table 3)  in terms of higher in-hospital mortality (7.4%, CI 6.6-8.3)  compared to the 2-4hr group (6.3%, CI 5.6-7.0), and 30 day mortality (12.5% for the under 2 hr group, compared to 10.9% for the 2-4 hr group).  The authors are perplexed by this — it doesn’t fit with their model that pneumonia’s progressive injury to the lung can be interrupted by antibiotics. They recommend further study (because recommending to withhold antibiotics until two hours had elapsed might be controversial).

A few more points about all the ‘benefits’ of early antibiotics  – JCAHO’s elevation of the 4 hr rule as a proxy for quality has encouraged hospitals to agressively, but superficially, treat coughing patients that walks into the ED. It’s hard to study whether this rule is contributing to antibiotic resistance, or whether rushing to give PO biaxin is really helping anyone, or whether the community benefits when we let pneumonia patients jump the triage queue at the expense of possibly sicker patients.

For further reading, emcrit.org has good resources on CAP, the blood culture controversy, and citations for optimal coverage for all of pneumonia’s different flavors.

Also, Jack previewed the new guidelines from the Infections Disease Society of America / American Thoracic Society, reprinted from Clinical Infectious Disease 2007: 44:S27-72:  (I love how the first author is, in fact, based in Canada.)

29. For patients admitted through the ED, the first antibiotic dose should be administered while still in the ED. (Moderate recommendation; level III evidence.)

Surprisingly mild!! The authors elaborate:

Blockbuster stuff! Let’s see if JCAHO modifies their recommendations in light of this authoritative, evidence-based recommendation.

Posted on Sunday, March 18th, 2007 at 5:25 am by Nick. Filed under Pneumonia, Regulations, Infectious Disease, Journal Club.
You may post a comment.