Sticking Blood Cultures and "HospitalCompare"


Nice study in the Journal of Emergency Medicine by Shapiro et al. showing that many of the blood cultures performed in the emergency department are low-yield.

At one institution the study was able to decrease the number of blood cultures by 27% and therefore decrease the health costs by nearly $125,000.

At this single study institution’s emergency department, there were 3901 blood cultures drawn over a 1-year period. The study analyzed patient characteristics and determined which patient characteristics were associated with a greater likelihood of bacteremia or “bloodstream infections.”

The symptoms more likely to be associated with bacteremia were divided into “major” and “minor” criteria.
Major criteria included temperature > 39.5°C (103.0°F), indwelling vascular catheter, or clinical suspicion of endocarditis.
Minor criteria included temperature 38.3-39.4°C (101-102.9°F), age > 65 years, chills, vomiting, hypotension (systolic blood pressure < 90 mm Hg), neutrophil% > 80, white blood cell count > 18 k, bands > 5%, platelets < 150 k, and creatinine > 2.0.

If a patient has either one major criterion or two minor criteria, then blood cultures were indicated. Otherwise, patients were considered “low risk” and did not need blood cultures.

Using study criteria, the researchers were able to reduce the number of blood cultures by 27%, resulting in  approximately 1053 fewer cultures per year. At an estimated cost of $15.91 per culture and a charge of $118 per culture set, there was a potential savings of $16,758 in costs and $124,286 in charges at ONE HOSPITAL.

In addition to preventing low yield blood cultures, the study also noted additional cost savings. Nearly 5% of patients in the study had false positive cultures, meaning that the cultures grew out an organism when there really wasn’t an organism present. In other words, the samples were accidentally contaminated. The study cited another study showing that patients with contaminated blood cultures have a hospital stay that is an average of 4.5 days longer and costs an average of $4385 more.

Contamination is one of those problems that it is difficult to guard against. Skin is cleansed with iodine and alcohol, but you can’t eradicate every single organism on the skin that could contaminate the needle. The more cultures you draw, statistically speaking, the more contaminated specimens you are going to get.

This study enrolled 3730 patients. Out of those 3730, only about 8% had true bacteremia and more than half that many – almost 5% – had false positives.

Using the study criteria, the researchers were able to identify more than 99% of patients who had positive bloodstream infections. Only 3 of the 3730 patients studied had infections in the blood that were not caught by using these criteria.

Yet, the government requires that we draw blood cultures before starting antibiotics on every patient that might have pneumonia.

I previously commented about how the CMS “Hospital Compare” website is a bunch of hooey.

It doesn’t allow people to compare government hospitals with all the other hospitals.
It purports to show which hospitals meet “quality” measures by performing certain tasks in a timely manner, yet many of the indicators it uses have little scientific basis.

The whole blood culture requirement is just one of the site’s many big failures.

Performing blood cultures before giving antibiotics in pneumonia patients has absolutely no effect on clinical outcome.
False positive cultures increase the length of stay by 4.5 days and increase the cost of hospitalization by nearly $5000.
Now researchers have come up with a way to decrease the number of blood cultures performed in the emergency department by 27%.

Yet, the Department of Health and Human Services and its “Hospital Compare” web site just keep the “stay the course” attitude. Looking for those “bugs of mass destruction” when finding them costs millions of dollars and has no effect on clinical outcome.

I’m sure that not all blood cultures are done on pneumonia patients. If we say that half of blood cultures were done for pneumonias, we would be able to save an average of $60,000 in costs in each of the 5,708 hospitals in the US … actually … lets just round down to 4,000 hospitals so that we underestimate costs … we’d save $240 million in costs.

Some of the ill-conceived indicators on the Hospital Compare web site are causing laypeople and hospital administrators to pressure physicians into practicing bad medicine. The NEJOM study is just one more reason why neither administrators nor the lay public should put much credence in the HospitalCompare statistics.

Just think, we’d save almost a quarter billion dollars per year by ignoring just one inaccurate statistic.

How much could we save if we just took the whole “Hospital Compare” site off the web?


  1. It is my understanding that you can just do one culture as long as you take at least 15cc of blood instead of a tiny splash that just darkens the bottle. The chance of contamination is reduced.

    Interesting idea. I tried to find something about this online and was only able to come up with a BMJ study in neonates.

  2. There was a potential savings of $16,758 in costs and $124,286 in charges at ONE HOSPITAL.

    So, you want to put the hospital out of business? These tests are a moneymaker. And the false positives are worth another $200 in profit for each test given (5% x $5000 at an 8:1 charge:cost ratio). The best criteria should be to give the tests (1) where indicated, and (2) if insured. (Yes, sarcasm)

    You gave me a WTF moment … until I saw the “yes, sarcasm” remark.
    Damn you. My brain can’t take shocks like that.

  3. Whitecoat, I 100% agree with you and this study. I read it last night and was hoping someone would do a writeup on it. So let me just say thank you.

    Likewise, I would like to hear what Happy Hospitalist has to say about this. Hopefully he will chime in, or any other IM guys lurking around.

  4. Many an idle moment has been spent waiting for these blood cultures to be drawn successfully so I can give antibiotics. It’s hard to get like 40 cc of blood (10cc per bld cx + 20 cc for the six million other sepsis labs) out of someone with no veins. A good way to cause a net delay in antibiotic treatment is to order blood cultures on everyone.

    PS Evidence-based practice has no place in JCAHO mandates. DUH.

  5. i’m confused…we have to draw blood cultures on all our pneumonia patients for ebm or cap protocol or some other mandated temp no elevated wbc…doesn’t matter. we admit from the er with dx like uti with sepsis…wtf..didn’t know cultures grew like in 2 hours. just a lowly floor nurse here but it does seem like overkill

  6. Pingback: Choosing a Doctor « WhiteCoat Rants

Leave A Reply