Mike the Mad Biologist

Do you want to know how to stop, or at least, lessen the next E. coli 0157:H7 outbreak? Improve our surveillance and public health infrastructure.

If we improve the infrastructure, we can speed the response time, making it easier to contain an outbreak.

Let’s walk through each of the steps the CDC outlined in its response.

1. Incubation time: The time from eating the contaminated food to the beginning of symptoms. For E. coli O157, this is typically 3-4 days.

There’s not much we can do to ‘improve’ this step. While random testing could be an option, when it comes to produce, I don’t see how inspections would detect bacterial contamination. The amount of produce nationwide that would need to be screened would be enormous.

2. Time to treatment: The time from the first symptom until the person seeks medical care, when a diarrhea sample is collected for laboratory testing. This time lag may be 1-5 days.

One thing that can improve seeking medical care is public outreach. If people know their symptoms should be taken seriously, then they are more likely to seek medical attention. Of course, having universal healthcare would help too…

3. Time to diagnosis: The time from when a person gives a sample to when E. coli O157 is obtained from it in a laboratory. This may be 1-3 days from the time the sample is received in the laboratory.

This is where improved infrastructure can really make a difference. Some hospital diagnostic laboratories are very well-equipped with state of the art equipment and trained personnel. Others, however, are not, and are overworked and underequipped. Funding for public facilities and cheapskate private systems (not all private systems are cheap) determines how quickly E. coli O157:H7 can be identified.

4. Sample shipping time: The time required to ship the E. coli O157 bacteria from the laboratory to the state public health authorities that will perform “DNA fingerprinting”. This may take 0-7 days depending on transportation arrangements within a state and the distance between the clinical laboratory and public health department.

There are several ways funding and infrastructure can reduce the time at this step. First, the communication technology between the clinic and the public health lab can be improved. For example, while Massachusetts is establishing a computerized surveillance network, currently, the technology revolves around faxes and phone calls (which means the response time can be slower if the report happens on a Friday night). County public health departments are even worse off. Critical time can be lost waiting for a response from and contacting public health departments. Second, to the shipping and transport times can be lowered with increased funding (either rapid shipping expenditures versus USPS, or couriers).

5. Time to “DNA fingerprinting”: The time required for the state public health authorities to perform “DNA fingerprinting” on the E. coli O157 and compare it with the outbreak pattern. Ideally this can be accomplished in 1 day. However, many public health laboratories have limited staff and space, and experience multiple emergencies at the same time. Thus, the process may take 1-4 days.

While a discussion of the inadequacies of the most common DNA fingerprinting technique, PFGE, would require a separate post, I think there are real problems with this technique: the training time is high, reproducibility isn’t very good, and the time to run the test is very long. Better technologies exist, but they would be expensive to implement: new machinery would be needed, and staff at the federal, state, and local levels would have to be retrained. The other reason more funding is need explains itself:

However, many public health laboratories have limited staff and space, and experience multiple emergencies at the same time. Thus, the process may take 1-4 days.

So here’s a few final thoughts:

1) Foodborne disease is relatively non-contagious. Imagine if this were something highly contagious, like influenza. Sleep tight.

2) Funding, funding, funding. All of these fixes require money. Expressing outrage isn’t enough-you have to ante up and kick in: it’s called taxation. To steal Oliver Wendell Holmes’ phrase, taxation is the price we pay for an outbreak-free society.

3) Biosecurity and biopreparedness require a public health infrastructure. Detecting an anthrax outbreak (and other bioterrorist agents) requires a public health system: the victims aren’t going to show up at FBI headquarters or army bases, they are going to be seen at hospitals and clinics. Rapid communication and diagnosis is essential. Biosecurity needs to be seen as an ‘add-on’ to a functional healthcare system. It can not be viewed as a parallel response system.


  1. #1 Karen the infectious disease epidemiologist
    September 25, 2006

    Great post! Always happy to see the importance of infectious disease surveillance touted. A few thoughts:

    Secondary spread via faecal-oral transmission is an important consideration for many enteric pathogens, including E. coli. The infectious dose of E. coli O157:H7 is very low; so while not as contagious as influenza, person-to-person spread is still significant. Appropriate hand hygiene is the single most important thing one can do to prevent infection with enteric and respiratory pathogens (but please forget about the antibacterial madness and just wash your hands frequently and properly for 20 seconds with plain old soap and water!).

    Perhaps it’s my Canadian perspective, but I am perplexed and annoyed by the American tendency to link Public Health to bioterrorism/biosecurity. Tying a crucial public service to a political whim makes those of us in the field nervous!

  2. #2 Sandra Porter
    September 25, 2006

    While a discussion of the inadequacies of the most common DNA fingerprinting technique, PFGE, [WIKI LINK] would require a separate post, I think there are real problems with this technique: the training time is high, reproducibility isn’t very good, and the time to run the test is very long.

    Totally true. I think most people, today, use PCR with specific primers and DNA sequencing. And this can be done pretty quick (under a day).

  3. #3 David Harmon
    September 26, 2006

    So, it sounds like a the major improvements would be at the outer range of response times. Funding and infrastructure could shave a day or so off the minimum time of 6 days, but the maximum of 22 days would be slashed to 10. Outreach and education could then cut perhaps 3 more days. (Figuring second-day response.) Of course, the facilities would be needed first, to handle the response to the outreach!

  4. #4 ScottG
    September 26, 2006

    My son contracted E. coli O157:H7 a couple of years ago, so this entry certainly piqued my interest. From the point of view of the patient (and their families), as well as public health, the time taken to identify the particular strain should obviously be as short as possible. When our doctor told us that it would be several days before the results were back (and this was at a good children’s hospital in NYC), I asked why they did not simply type by PCR. He sent me references (which I do not have here) that suggested that PCR was less reproducible than PFGE (even though PFGE does seem like an archaic technique- but if it works?). These links seem to support this:

    The second reference concerns optimizing a PCR-based typing technique. Although this is a relatively automated process, training is still required- and PCR is such an unbelievably sensitive technique that the slightest error will show up. For a relatively complex PCR, as is described here, the PCR machines and all reagents will need to be standardised. Not as easy as it first looks.

    I agree that more money is needed- both for quicker, better staffed testing and also for more research into reproducible, accurate testing methods.



  5. #5 Paul Orwin
    September 26, 2006

    I think an AFLP type of approach would work well for getting very fast, highly repeatable sampling. Also, since there has been some comparative genomics done looking at O157 v. K-12 (lab E. coli), I suspect a specific PCR to id the presence of O157 in stool could be done is a few hrs, rather than a day. quantitative and diagnostic PCR should be reasonable for a well trained tech in a decent clinical lab.

  6. #6 ScottG
    September 26, 2006

    I have had a glance at a few AFLP papers- this PCR comparative study, for instance: Clinical Microbiology & Infection
    Volume 9 Page 823 – August 2003. If this paper represents a typical protocol, then I think it is unlikely to be used in its present form as a reproducible diagnostic assay. It is simply too complex, with too much room for error. Although I have not performed AFLP, I have significant experience with another PCR based assay, the TRAP assay for telomerase activity. This type of assay does not translate well from the research lab to a routine, clinical setting. The data from the above cited paper, although in favor of AFLP, had an error rate of over 15% (1 of 6 reproduced assays gave a non-consistent result), and would take up to two days to perform.