A Medical-Skeptical Classic

The medical literature slowly becomes outdated. As a result there are not that many ‘classics’ in the field, since their content becomes less relevant. The medical aphorism is that 10 years after graduation from medical school, half of everything you learned will no longer be valid. The problem for medical students is trying to figure out which half of their curriculum is not worth learning.

Old studies become increasingly irrelevant as diagnosis and treatment changes over time under the relentless pressure of medicine. I once came across the best of Osler, with his descriptions of typhoid fever and pneumococcal pneumonia. The essays were far more literary in style than today’s journal articles, describing the presentation of these diseases in Dickens-like detail, but of little practical help given the advances in treatment and the understanding of the microbiology of diseases.

Technology also expands and limits what papers are available. If there is not an electronic form of an article, it might as well not exist. Many classic articles are not yet available in digital form, and the article in question for this post I had to get as a scanned version of the original paper, rather than a pdf. As a result of time and lack of electronic access, much of the older medical literature is not easily accessible, and journal publishers are not particularly interested in the free dissemination of information. Which is a shame. There is the occasional older reference that is as applicable today as when it was published.

There was never, to the best of my recollection, a time when I was not a skeptic. But there was a time when I had neither the time nor the knowledge to be able to think skeptically about medicine. The torrent of information that needs to be assimilated in medical school, residency and the first part of a fellowship makes reflection about that information almost impossible.

I do remember an article that was a turning point in my thinking about thinking about the practice of medicine. It was from 1989, the last year of my Fellowship, and was published in the American Journal of Medicine, entitled “Observations on spiraling empiricism: its causes, allure, and perils, with particular reference to antibiotic therapy” by Kim and Gallis (hence forth called OOSE, pronounced “ooze”, I suppose).

If you can scrounge up a copy, by all means do so, as it is a classic. It is a collection of logical fallacies and critical thinking as applied to infectious diseases. It was the first time I read an article that discussed how to think about thinking in medicine. I had no idea that there were logical fallacies. Most of what passed for critical thinking in my training concerned understanding the statistics, materials, and methods of studies. Important, but limited.

OOSE starts with a description of the discovery of antibiotics and the amazement of physicians that for the first time diseases that were often fatal were now curable. It must have been an amazing time for physicians when infections that routinely killed were suddenly vanquished. I have witnessed a similar revolution with the advent of HAART, where AIDS has gone from a nine-month life expectancy to a chronic disease with perhaps a normal life expectancy. At my hospital is Dr. Charles Grossman, now in his 90′s and still attending conferences. His long and productive medical career started at Boston Hospital where, as an intern, he was involved in giving the first dose of IV penicillin to a lady who was dying of a streptococcal infection, who survived another 50 years.2

Antibiotics have been developed to kill increasingly resistant and virulent bacteria, and until recently we have managed to keep one step ahead of the organisms. Unfortunately, the ability to become resistant is outstripping out ability to develop new agents and we are slowly, and inexorably, losing the battle, sliding in to the post antibiotic era.

Part of the evolution of resistance occurs from the inappropriate use antibiotics, which can often be due to uncertainly of the diagnosis. But inappropriate antibiotic use can also due to faulty thinking.

As the authors of OOSE note, “The imprecision of clinical practice establishes context, the litigious nature of our society unnerves; the absence of toxicity permits; and the sum of these encourages the incontinent, extemporaneous use of these antimicrobial agent.”

Incontinent use of antibiotics leads to increasing bacterial resistance, and the use of antibiotics when they have not been needed has accelerated the evolution of pathogens, occasionally to the point where there are infections I cannot cure and bacteria that I cannot kill.

The term spiraling empiricism describes the inappropriate treatment, or the unjustifiable escalation of treatment, of suspected but undocumented infectious diseases. Empiricism and empirical therapy, defined as the carefully considered, presumptive treatment of disease prior to the establishment of a diagnosis, often are necessary in the proper practice of medicine. On the other hand, ill-considered or inappropriate use of antibiotics, incurring unnecessary risk and expense, should be indicted and condemned. The difficulty lies in distinguishing reasonable or appropriate from unreasonable or inappropriate therapy.

As a teaching physician in a teaching hospital, I notice that sometimes it is the FUD, Fear Uncertainty and Doubt, combined with faulty thinking that sometimes leads to the inappropriate use of antibiotics.

OOSE provide a conceptual framework for approaching diseases and potential therapy (see table below). Observation, prophylaxis, empirical therapy, therapeutic trial, and specific therapy. Of these the first and the last are, sadly, the least used. For interns and resident, the motto is ‘Don’t just stand there, do something’, and with the pressures to shorten a hospital stay as much as possible, simply watching the patient is a luxury few can afford. As an experienced physician, I feel much more comfortable with the motto, ‘Don’t just do something, stand there’, or as the paper calls it “masterly inactivity.” As my wife can attest, I am the master of doing nothing.

Specific therapy in infectious diseases is not as common as I would like given the vagaries of growing the infecting organisms and the degree to which one wants to maximize diagnostic certainty. I could probably get the etiology of every pneumonia admitted to the hospital with an open lung biopsy, but it would hardly be worth the resultant morbidity and mortality.

Most of the time the patient is ill enough to be admitted to the hospital and, after appropriate studies and cultures are done, empiric therapy is started. That is often not an unreasonable course of action. These days you have to be ill to get admitted and it is the rare patient who comes into the hospital who can wait for cultures to be positive before beginning antibiotics. I cannot emphasize enough how ill patients are when they are admitted to the hospital, and how unclear the proper diagnosis can be at the beginning of a hospitalization. After a day or two all the diagnostic information has returned and, with the 20:20 vision of hindsight, the correct diagnosis may become clear and as a result the proper course of therapy is clarified.

Since cultures are often negative, the empiric course of therapy may morph into a therapeutic trial.

After setting the therapeutic framework in place, OOSE proceeds, with case reports, to describe fallacies in antibiotic therapy that lead to the wrong therapeutic interventions. There is, in medicine, a long tradition of using cases as illustrative of problems, but not as anecdotal evidence for the proof of a hypothesis. We all remember concepts when they are applied to specific patients and specific cases.

Their Fallacies in Antibiotic Therapy are

  • Broader is better.
  • Failure to respond is failure to cover.
  • When in doubt, change or add another.
  • Sickness requires immediate treatment.
  • Response implies diagnosis.
  • Bigger disease, bigger drugs.
  • Bigger disease, newer drugs.
  • Antibiotics are non-toxic.

I will add my own.

  • Once started, an antibiotic cannot be stopped.
  • Once a class of antibiotic is started, you need to stay in class.
  • Gotta double cover (i.e. give two antibiotics) a particular organism.
  • The primary reason a particular antibiotic is given is “I like it.”

I will not repeat the OOSE case studies as examples to support the fallacies, but instead will make comments and substitute my own examples. These fallacies are used a minority of the time, but it is not common to explicitly discuss the ways in which medical reasoning can go wrong in the context of faulty thinking. These fallacies are not unique to infectious diseases, and I am sure a neurologist or oncologic surgeon could generate a similar list in their own specialty. I doubt, however, that the same could be said of a chiropractor, naturopath, or homeopath. Perhaps one will prove me wrong in the comments, but logical fallacies are the sine qua non of alternative medicine. They are parallels and subsets of the standard logical fallacies beloved by skeptics everywhere. They are examples of the endless number of ways we can all think inaccurately.

Broader is better

Broad spectrum means an antibiotic that covers every conceivable pathogen. For most infections, the potential causes are often few in number, based on the history and potential exposures. You will always be surprised by the unexpected positive culture and any therapy will have gaps.

Residents do not want to be accused of not covering some particular organism if it turns out that one of the patients with an odd organism growing in their blood. So they give antibiotics that kill everything, adding expense and driving resistance without necessarily improving care.

Failure to respond is failure to cover

Few infections get better immediately, and some will be symptomatic for days. Staphylococcal endocarditis will have fever for up to 2 weeks, and often the reason a patient does not respond is not because they are on the wrong or inadequate antibiotic but it is either the natural history of the disease under treatment or, more often, there is a need to drain some pus. The old aphorisms “Nothing heals like cold steel” and “A chance to cut is a chance to cure,” remain as true today as they did in the pre antibiotic area. One of my colleagues does an impression of me when I am 85 and demented in a nursing home, banging my hand on the table, shouting over and over “drain the pus.”

Adding antibiotics may add only cost, potential toxicity and accelerated resistance.

When in doubt, change or add another antibiotic

Medicine is filled with uncertainty, and often it is the case that if an infection is thought of, it is treated, no matter how unlikely it is that it may be causing the disease. I have a fantasy where oncology is practiced like infectious disease. “It might be lymphoma, so let’s start with CHOP, but we don’t want to miss adenocarcinoma, so let’s add bleomycin, and in case it’s breast cancer we need to include tamoxifen and it could be prostate so let’s add…etc , etc.”

When in doubt, increase your diagnostic certainty.

Sickness requires immediate treatment

Sometimes it does, sometimes it doesn’t. It depends on the patient and how clinically stable they are. All too often, people receive therapy for diseases that do not need antibiotics or the antibiotics are given before a firm diagnosis is made. This is driven, in part, by CMS core measurements. Community-acquired pneumonia guidelines mandated that antibiotics be given within 4 hours of the presentation to the ER, if not the hospital gets a demerit. The result? Perhaps one in five patients who get treatment for pneumonia did not turn out to have pneumonia and would have benefited from watching and waiting.3 No good deed ever goes unpunished.

Healthcare providers appear to be particularly hesitant to not treat a fever with antibiotics. I like to say, antibiotics are not antipyretics, if you want to treat a fever, give Tylenol.

Response implies diagnosis

This is the most difficult fallacy for people to abandon. Patient has a fever, no diagnosis is evident, but the fever went away at some point after the institution of antibiotics.

Most fevers go away. Many diseases that are not infectious will have a fever. This is the medical equivalent of the skeptical motto ‘association is not causation’. The worst cognitive error physicians and patients make is the assumption that in the absence of a good diagnosis, the improvement when a therapy is given is due to the therapy. Rigorously and consciously avoiding this error is key to being a good healthcare provider.

Bigger disease, bigger drugs

MRSA is a current bane of infectious diseases. It is a bug that I cannot always kill with the current antibiotics, and the three or four antibiotics to which the organism is susceptible are all equally bad. So, when there is a severe MRSA pneumonia, the urge is to add rifampin, even though the data to support its use is problematic at best. But if one drug is good, are not two better? Perhaps. Sometimes combining antibiotics may be antagonistic and decrease the efficacy of therapy.

Bigger disease, newer drugs

Of course newer is better. New and improved. It is as American as Procter and Gamble. Not necessarily. For MRSA, there are no better drugs than nafcillin or cefazolin, and the newer agents are less efficacious in head-to-head trials. However, newer drugs are heavily advertised and detailed, <sarcasm> not that any physician would ever be swayed by advertisements instead of data </sarcasm>.

Antibiotics are non-toxic

All drugs have side effects, some common, some rare. Certainly antibiotics are less toxic than chemotherapy agents, but they do have the potential to cause significant harm. We all have our irrational beliefs. Here is mine (besides believing in critical thinking): The chance of an adverse reaction is inversely related to the need for the antibiotic. So if the patient has little need for the antibiotic, then the chance of an adverse effect is almost 100%. I know that it is not true, but some days the universe sure seems to conspire to function as if it were.

Some antibiotics are a big gun, are strong, or powerful

There are few things in medicine with 100% sensitivity and specificity. However, if your healthcare provider uses the adjectives “big gun”, “strong”, or “powerful” in reference to an antibiotic, they are either 1) talking out their backside or 2) ignorant about antibiotic use. 100% sensitive and specific.

If I had neurosyphilis, and there are those who suspect I do, the ‘strong’ or ‘powerful’ ciprofloxicin would do nothing to treat my infection, but ancient, weak old penicillin remains the treatment of choice. What you want to give are appropriate antibiotics: something that will reliably kill the organisms in whatever space is infected. These adjectives are advertising ploys used on fool gullible rubes, er, I mean healthcare providers, to think they are doing what is best for their patient. They provide a false security that you are giving the patient the best therapy.

Once started, an antibiotic cannot be stopped

People have trouble stopping an antibiotic, even when the cultures come back suggesting a particular therapy is not needed. A patient comes in septic and antibiotics are begun. Even when it turns out that the blood grows a penicillin-sensitive S. pneumoniae, the multiple initial empiric antibiotics may be continued. One of my colleagues only half-jokingly says the reason she consults me is to have me stop the antibiotics. When there is reasonable data to suggest that shorter courses of antibiotics are no better than longer, the endless allure of 10, the number of fingers, or 14, twice two weeks, is often the default duration of therapy. I once saw a patient who had been on IV vancomycin for 6 months for osteomyelitis of the foot; the provider just couldn’t bring themselves to stop the therapy.

Once a class of antibiotic is started, you need to stay in class

If a patient is admitted with an infection and started on an intravenous quinolone and grows something than can be treated with a less expensive, narrower spectrum agent like amoxicillin, at discharge the patient is likely to go home on expensive oral quinolone. They responded to a class of antibiotics, and so it needs to be continued, even if the cultures say otherwise. Magical thinking, as best I can tell, and one that I occasionally participate in.

Gotta double cover (i.e. give two antibiotics) a particular organism

There is a persistent myth that some infections, like Pseudomonas, always require two antibiotics. That is true occasionally, but for most cases a single antibacterial is no worse than two, and the more antibiotics you give the more you increase the toxicity and the faster you drive antibiotic resistance.

The primary reason a particular antibiotic is given is that “I like it”

I like to say that the three most dangerous words in medicine, especially when it comes to therapeutic interventions, are “In my experience.” In my experience you can’t trust anyone who uses that phrase. Remembering hits and forgetting misses drives antibiotic use more than I would like to admit. Infectious disease docs are sometimes the opposite. We put too much emphasis on the antibiotics that have failed us in the past. It is one extreme or the other.

Patient is admitted with a complicated infection and started on X. Why X? I ask. It worked for me in the past is the reply. What are you trying to kill? I might then ask. Often, they say the boogie man of the ICU, “Pseudomonas”. What, I will continue, is the chance drug X will be effective against Pseudomonas? They don’t know. So why again are you using X? It is what we do. They have used it successfully in the past, so it should work again.


Fortunately, most drugs work most of the time in most patients, but that rule is slowly being lost as the organisms become increasingly-resistant to our current armamentarium of antibiotics and there are few, if any, replacements in the pipeline. At my institution there is 2% resistance for Pseudomonas to ceftazadime and 4% to pipercillin/tazobactam. We get a few cases a year at best of a bloodstream infection and sepsis from Pseudomonas. Most physicians are not going to see enough infections by a given organism to get a sense of what does and does not work. It is why you cannot trust your experience.

I have not yet read the book “How Doctors Think” although being in a teaching hospital I try, however poorly, to demonstrate how I think about thinking about infections. I often give the residents on service a copy of the article on which this post is based. Once, when I was ranting about this at a conference, one of the attendees said I was an arrogant subspecialty who didn’t understand the pressures faced by general internists.

I don’t know. Perhaps they are right. But often getting the right diagnosis and therapy is less about what you know and more about being rigorous about understanding how you know. Only when you are conscious of your ability to think poorly, can you compensate. In the era of Google, any knowledge gap is a search away, but you have to be aware of the knowledge gap first.

To finish quoting this under-read and under-appreciated article:

…fashion and marketing forces, not science, shape clinical practice. Within an intensive care unit, gravity of illness amplifies the effects of delay and presses for preemptive, urgent empiricism, but the selection and manipulation of therapy can sometimes be justly criticized. The intensive care unit mentality, which treats all illness as life-threatening, favors therapy over careful consideration of diagnostic maneuvers…The potential for, and tendency toward, spiraling (as opposed to appropriate) empiricism appears disturbingly pervasive – reinforced by fashion, compensation, and litigation.

True, but driven more by the urge to help people get better and the fear of doing harm by missing an infection. OOSE does overstate the problem, but such is the nature of polemics.

Osler’s comment that the ‘practice of medicine is an art, based on science’ underscores the dichotomy and the dilemma, of medical science. Empiric therapy at its best is set between the Scylla of unnecessary delay and the Charybdis of therapeutic voyeurism. Knowledge of the science of medicine and the natural history of disease should temper and compliment the art.

This article was one of the few epiphanies in my medical career. I re-read it several times a year and consciously endeavor to keep its concepts in the front of my thought processes. It stays fresh with each reading, because my capacity to think badly is limitless and only by constantly reminding myself of that fact can I hope not to stray into illogic.

Part of applying the art of medicine is to recognize the fallacies of medical thinking and to question your decisions and interventions in the light of understanding yourself and how you think.

The bumper sticker “Don’t believe everything you think” should be a motto for all health care providers.


  1. Kim JH, Gallis HA. (1989). “Observations on spiraling empiricism: its causes, allure, and perils, with particular reference to antibiotic therapy.” Am J Med, 87(2):201-6. PMID: 2667357
  2. Grossman CM. (2008). “The first use of penicillin in the United States.” Ann Intern Med, 149(2):135-6. PMID: 18626052
  3. Wachter RM, Flanders SA, Fee C, Pronovost PJ. (2008). “Public Reporting of Antibiotic Timing in Patients with Pneumonia: Lessons from a Flawed Performance Measure.” Ann Intern Med, 149:29-32. PMID: 18591635

Posted in: Clinical Trials, Science and Medicine

Leave a Comment (20) ↓

20 thoughts on “A Medical-Skeptical Classic

  1. Jules says:

    Very nice article! One of the best I’ve read here.

    Now if only such a thing existed for environmentalism….

  2. David Gorski says:

    All drugs have side effects, some common, some rare. Certainly antibiotics are less toxic than cancer chemotherapy agents, but they do have to potential to cause significant harm. We all have our irrational beliefs. Here is mine. Besides believing in critical thinking. The chance of an adverse reaction in inversely related to the need for the antibiotic. So if the patient has little need for the antibiotic, then the chance of an adverse effect is almost 100%. I know that it is not true, but some days the universe sure seems to conspire to function as if it were.

    The same seems to be true of surgery. The “softer” the indications for an operation are, it seems, the higher the likelihood of significant complications from that operation. At least, as you put it, the universe seems to function that way. :-)

    Yes, I realize it’s probably selective memory and confirmation bias on my part.

  3. Diane Henry says:

    If anyone can figure out a way to get a copy, would you post? I’ve asked my public library to email me a copy via interlibrary loan, but am still waiting to receive it (it’s been a couple of weeks–Mark mentioned this in one of the Quackcasts). Thanks!

  4. weing says:

    I was able to download it from my son’s medical school library. I’m pretty sure that the hospital you are affiliated is able to get the article for you.

  5. weing says:

    re the irrational beliefs, I think we are all susceptible to them. The less any medication is indicated, the more likely the side effects is one of them. Trying to rationalize it, I keep thinking of inversion of the risk benefit ratio of any treatment. An excellent book about this tendency:
    How We Know what Isn’t So
    By Thomas Gilovich

  6. caoimh says:

    Hi Mark,

    Great article.
    I’m not in the medical field myself, surprisingly however many of the comments you made work quite well when I apply them to my own area.



  7. Diane Henry says:

    I’m not in the medical profession, but I’ll definitely check the med school library. Hadn’t thought of that. Thanks.

  8. wch says:

    “Empiric therapy at its best is set between the Scylla of unnecessary delay and the Charybdis of therapeutic voyeurism.”

    Now that’s a kind of flair that you don’t often see in the scientific literature. Sexy.

    Thomas Kida’s “Don’t Believe Everything You Think” was next in my reading queue, but I think this article just jumped the line.

  9. skeptyk says:

    I have been hearing of, and reading titillating excerpts from, this article for years, and appreciate your commentary on it, particularly personally cogent this week in a conversation about antibiotics. I think I’ll try the interlibrary loan system. My son is in community college and all the state schools and UVM, which has a med school, are linked in, IIRC.

  10. The article is great, and so is the post. Mark, your blend of clinical wisdom and sardonicism is all too rare in medical training contexts. One comment, not really a disagreement:

    Most people, and probably even some SBM readers, don’t understand the general basis for the emergence of antibiotic resistance. Overuse of antibiotics doesn’t encourage individual microorganisms to develop molecular mechanisms to resist those antibiotics. Rather (evolution 101), genes for such mechanisms already exist in a tiny percentage of most populations of those microorganisms, or at least in genetic reservoirs that are readily available to those organisms(evolution 201, which we needn’t go into), and by giving an antibiotic that kills most of the sensitive organisms we have conferred a selection advantage on the resistant ones–which were heretofore at a disadvantage (that’s why there are so few of them) if only because carrying unnecessary genetic baggage exacts a metabolic price.

    Simplified example: Imagine a species of pathogenic (disease-causing) bacteria in which, for typical populations, 1 in a million individuals is resistant to penicillin. At first, most courses of treatment are successful, because the vast majority of the population in question is killed by the drug and the rest are either killed by immune responses, both specific and non-specific, or revert to “carrier” status if the organism and the host have that kind of a relationship. Even most carrier-status organisms will, for a while, continue to be those that are sensitive to penicillin, simply because their numbers had so overwhelmed the resistant ones to begin with. At first, moreover, new opportunities for such populations to make trouble will continue to be dominated by penicillin-sensitive organisms (assuming that the host is not taking penicillin chronically–which some are, hint hint) because those sensitive organisms remain better competitors in general (the metabolic price thing).

    But at some point, after many courses of penicillin in many hosts, and after many examples of penicillin being given to humans and animals chronically, the resistant organisms will inevitably gain the upper hand of “survival of the fittest.” For that matter, for evolution 201 reasons, they may also have become more robust (than their own ancestors) in other ways. Our ministrations have not literally created or encouraged the formation of these individual organisms–rather, we have selected for them. We have encouraged their proliferation.

    Now, consider the “double cover” fallacy. I know that Mark meant something a little different from what we’re talking about here, but clarification is needed. Most “wild” populations of microorgisms have several different antibiotic resistances scattered among individuals. Prior to our meddling, the co-existence of any number of such resistances in any single member of the population was strictly a matter of chance (antibiotics, and hence resistances, have many different mechanisms of action). If the frequency of each resistance was 1 in a million (a high incidence, by the way, prior to the era of meddling), the chance that a single organism had three was therefore 1 in 10^18. Since there are only 10^13 bacteria/human, the chance that triple-antibiotic coverage of an infection will select for antibiotic-resistance seems vanishingly small.

    Understand that this is simplified: it says nothing about additive toxicities of antibiotics, but more importantly it ignores the difference between impregnating the agar in a petrie dish with 3 different antibiotics and treating a person with the same three. In the latter instance, there are bound to be variations in tissue levels, compliance (ie, outpatients may not take the pills according to the intended schedule), antibiotic effects are not necessarily all-or-none, and other, murkier things. Any of these perturbations could select for resistance to one of the antibiotics, which would eventually cut our multiple coverage advantage by a factor of a million, and so forth. This, alas, already happens in the real world.

    Nevertheless, there are times when triple (not double, usually: 1 in 10^12 is too goddamned high) coverage is better, precisely because it “discourages” selection of resistant organisms. Mark knows the field far better than I, but for years the standard treatments of TB (a disease that is now coming back with a vengeance because of antibiotic resistance) and of HIV (not a bacterium, but subject to the same genetic issues) have required multiple, simultaneously administered drugs, mainly to guard against the blossoming of resistant strains. It could be argued (I bet it has been, but I don’t follow the literature) that there might be little antibiotic resistance among tubercle bacilli if triple antibiotic use had been the norm from the beginning (not that it could have been, because effective anti-TB drugs were discovered at different times).

    Please don’t anyone misconstrue this comment: I’m not arguing for multiple coverage as a general policy. Also, I know that there are plenty of technical objections, exceptions, and clarifications that could be brought to bear on this simplified discussion. It just seemed to go along with this post.

  11. daedalus2u says:

    I would add another fallacy; that the only adverse side effects are due to metabolic effects in the host. Destruction of commensal organisms is probably a more serious adverse side effect, but one that is harder to study because it is more idiosyncratic (it depends on the non-pathogenic bacteria that were there to begin with and how the antibiotic affects them).

    A very interesting article on how the diversity of bacteria changes during treatment with antibiotics is here.

    My interpretation is that a very large part of what is keeping the initial infection in check is non-pathogenic bacteria. When those non-pathogens are wiped out by the antibiotic, the pathogen has an open niche and can expand rapidly, even though it is susceptible to the antibiotic being given.

  12. Mark Crislip says:

    in pulmonary tb there are huge numbers of mycobacteria, some of which have mutated at known rates, so when you treat pulmonary tb with a single agent, you are selecting for pre existing resistant mutants.

    hiv is such a sloppy reproducer and multiplies at such a ferocious rate it is estimated that in a 24 period it undergoes three mutations at each site in the genome. i remain amazed that haart has worked as well as it has for so long, and I nervously await a breakthrough.

    neither the tb nor the hiv model may apply to bacterial infections.
    very broadly, antibiotic resistance comes in two flavors:

    pre existing resistance, usually due to inactivating enzymes and then there are acquired resistance, often due to single amino acid substitution of antibiotic binding sites.

    use of antibiotics should select for more mutants due to the latter mechanism.

    bacteria, when they are exposed to environmental stress increase their mutation rates in an attempt to ‘escape’ the stress by evolving their way out of it.

    when it comes to bacteria and antibiotics, you would expect that the more antibiotics there are in the environment, the faster you would get resistance, and when it comes to clinical bacterial infections the use of combination therapy has not reliably been shown to decrease the emergence of resistance. then this:

    Proc Natl Acad Sci U S A. 2008 Sep 16;105(37):13977-81. Epub 2008 Sep 8.
    Accelerated evolution of resistance in multidrug environments.
    Hegreness M, Shoresh N, Damian D, Hartl D, Kishony R.

    The emergence of resistance during multidrug chemotherapy impedes the treatment of many human diseases, including malaria, TB, HIV, and cancer. Although certain combination therapies have long been known to be more effective in curing patients than single drugs, the impact of such treatments on the evolution of drug resistance is unclear. In particular, very little is known about how the evolution of resistance is affected by the nature of the interactions–synergy or antagonism–between drugs. Here we directly measure the effect of various inhibitory and subinhibitory drug combinations on the rate of adaptation. We develop an automated assay for monitoring the parallel evolution of hundreds of Escherichia coli populations in a two-dimensional grid of drug gradients over many generations. We find a correlation between synergy and the rate of adaptation, whereby evolution in more synergistic drug combinations, typically preferred in clinical settings, is faster than evolution in antagonistic combinations. We also find that resistance to some synergistic combinations evolves faster than resistance to individual drugs. The accelerated evolution may be due to a larger selective advantage for resistance mutations in synergistic treatments. We describe a simple geometric model in which mutations conferring resistance to one drug of a synergistic pair prevent not only the inhibitory effect of that drug but also its enhancing effect on the other drug. Future study of the profound impact that synergy and other drug-pair properties can have on the rate of adaptation may suggest new treatment strategies for combating the spread of antibiotic resistance.

    PMID: 1877956

    which to my understanding makes far more sense.
    double coverage in the long run will increase the rates at which resistance will evolve in your hospital

  13. Mark,

    Thanks for the info, and it is interesting stuff. I haven’t been able to read the entire Hegreness paper (only the abstract is available online for free, and I’m away on vacation), but it strikes me that it might be concluding more than it can justify. I write this with some trepidation, because I’m way out of my league here, but it’s more fun than a crossword puzzle, so here goes.

    First, the title says “multidrug” but in several places the abstract suggests two drugs, not more, per assay. For reasons stated above, I don’t believe two-drug combinations are an adequate test of the question. Second, the finding that “mutations conferring resistance to one drug of a synergistic pair prevent not only the inhibitory effect of that drug but also its enhancing effect on the other drug” would only be surprising if it were not true, by definition. I know what “synergistic” means with regard to antibiotics, by the way, but I don’t know what “antagonistic” means. I’d have thought that antibiotics are either synergistic or merely have unrelated mechanisms. Perhaps the implication is that a drug that depends on attempted bacterial cell division, for example, is “antagonized” by a drug that poisons protein synthesis to the extent that the cell never attempts to divide?

    Third, you can’t tell from the abstract to what extent the conclusions are based on assays in which “subinhibitory drug combinations” were used. That condition, it seems to me, would be the most likely to result in stable strains of resistant organisms, since some bacterial growth would, by definition, proceed in each case. Thus it does not really test the interesting question. On the other hand, as mentioned above, I’ll grant that it’s reflective of some real world, clinical settings.

    Finally, I take issue with this sentence:

    “Although certain combination therapies have long been known to be more effective in curing patients than single drugs, the impact of such treatments on the evolution of drug resistance is unclear.”

    Wait just a damn minute: OK, it may be true that the impact of combination therapies on the evolution of drug resistance in general is unclear, especially because of the vicissitudes of the clinical as opposed to the laboratory setting, but isn’t it generally accepted that at least some of those “combination therapies have long been known to be more effective in curing patients than single drugs” precisely because they are not foiled by the emergence of resistant strains?

    Wasn’t that the difference, for example, between the tinidazole-only group and the tinidazole + bismuth group in Marshall and Warren’s first trial of treatment for H. pylori (5% and 70% eradication, respectively; Lancet 1988), and also explained the success of the triple drug (metronidazole/tetracycline/bismuth) regimen of Graham, Lew et al (Gastroenterology 1992), in which there was an 87% eradication rate overall, and a 96% rate among subjects “who took >60% of the prescribed medications”?

    One other point for the evolution-naive, which is the main reason that I wrote my first comment above. I know this is not you, Mark, but you write:

    “bacteria, when they are exposed to environmental stress increase their mutation rates in an attempt to ‘escape’ the stress by evolving their way out of it.”

    ‘We,’ by which I mean biologists and applied biologists such as ourselves, talk this way all the time because it is a handy shortcut. It ain’t really how things happen, however. Individual bacteria don’t increase their mutation rates in response to stress, nor does the increase in such rates found in populations of bacteria under stress reflect an “attempt” to do anything. Just as there are pre-existing bacterial resistances in some members of most bacterial populations, there are also pre-existing differences in mutation rates. That’s because DNA is being damaged all the time, but most of the damage is fixed by DNA repair enzymes (also true for us). Only the damage that remains after the repair will be manifest as a mutation. But there are polymorphisms in genes that code for DNA repair enzymes just as there are in other genes, so some repair enzymes work better than others.

    Under ‘non-stressful’ circumstances, bacteria with better DNA repair have a huge selective advantage, because most mutations are harmful-to-lethal. Thus the incidence of rapid mutators, ie poor DNA repairers, is normally kept to a small, frictional level (why isn’t it zero? because it itself is continually renewed by…mutations in the genes that code for DNA repair enzymes! Ain’t Bio great?).

    When antibiotics make things “stressful” however…I think the rest is obvious.

  14. The Blind Watchmaker says:

    “In the era of Google, any knowledge gap is a search away, but you have to be aware of the knowledge gap first.”

    I want the T-shirt.

    Antibiotics are often treated like CAM therapies (excuse me, SCAM therapies) in primary care offices. They act as a proxy for actually thinking about and explaining what is going on to the patient. I could see 2 or 3 patients and give them each amoxicillin in the time it would take to explain the nature of viral URI’s to frustrated patients.

    I could also code a level 3, or even a 4, for the visit if a prescription med is used. Advice is usually worth only a level 2 visit (3 if over 15 minutes is spent).

  15. Correction:

    “Just as there are pre-existing bacterial resistances in some members of most bacterial populations”

    I meant to write “Just as there are pre-existing ANTIBIOTIC resistances in some members of most bacterial populations”

  16. daedalus2u says:

    Just thinking out loud, but if selecting for resistant organisms is bad, should we instead try to foster non-resistant organisms? That is do things to make non-resistant organisms less rare? In other words encourage their growth? If you sprayed non-resistant organisms everywhere, that is going to dilute the resistant ones. You obviously can’t use potentially opportunistic pathogenic organisms where they might cause infections (as in operating rooms), but if they are non-pathogens they should have no adverse effects anywhere else.

    Many of the standard antimicrobial agents (pine oil, phenolics, triclosan, chlorhexidine, even quaternary amines) do select for efflux mediated resistance which can produce broad spectrum antimicrobial resistance.

    It seems to me that the goal should not be to try and reduce all bacteria populations to zero (which is not possible on living humans), but to reduce the likelihood of infection with something which cannot be treated as much as possible. This is a more complicated goal, because the number of bacteria in an infectious dose is not a constant, but depends on what else is present. A surface that has 10 infectious organisms where the infectious dose is 5 is more dangerous than a surface that has 100 of the same organism, but because of the presence of other bacteria the infectious dose is 500.

  17. @daedalus,

    A caveat about your comments: other than in the gut or vagina, most bacterial infections that require treatment are pure cultures of a single species that occur in parts of the body that are normally sterile, eg, lung, kidney, bladder, urethra, blood, bones, joints, heart valves, brain, or other soft tissue sites. In these cases, the goal IS and must be to reduce the bacterial population to zero, even if this is not always accomplished by antibiotics (for deep tissue abscesses, for example, surgical drainage is usually the treatment of choice; the body takes care of the rest). Except in immunocompromised or otherwise very ill patients, this is usually successful, although we worry about possible, multiple antibiotic resistances in the future, as Dr. Crislip discussed.

    Thus other organisms don’t pertain to the eradication of these infections, although you are correct that antibiotics almost always cause perturbations in the normal populations of commensals in the gut (defined as mouth to anus and including the biliary tract, but most importantly the colon) and vagina, which can lead to different problems.


  18. daedalus2u says:

    Dr. Atwood, I don’t disagree, but in the paper I linked to above, they were looking at infections of the lung, and they found very diverse populations in the patients when they initially presented. Normal lung in healthy controls was bacteria free.

    What I found quite interesting is that they did find the bacteria I am working with, ammonia oxidizing bacteria. These don’t have any virulence factors, so they can’t be “pathogens”.

    The dominant population was not Pseudomonas to start with, but Pseudomonas became dominant after treatment with antibiotics. I don’t know how many of the identified bacteria are opportunistic pathogens, but the ones I am working with are obligate autotrophs and so I don’t think can be pathogens of any sort.

    Pseudomonas does use NO as a signaling molecule, and a major virulence factor for Pseudomonas is formation of a biofilm. NO blocks that and disrupts biofilms that already exist. Inhibiting formation of a biofilm may prevent Pseudomonas from becoming dominant.

  19. daedalus,

    Good point, and I should have mentioned that sort of exception: chronic infections or colonizations of anatomical sites that are normally sterile, due to prolonged violations of normal mechanical defenses against such colonization–such as endotracheal tubes, as in the paper you cited. Another relatively common example is in someone with frequent or continuous instrumentation of the urinary tract, such as after a spinal cord injury. In these patients, as you point out, it is inevitable that such sites will become colonized by many different organisms.

  20. daedalus2u says:

    Again, I don’t disagree, but I don’t think that there has been much looking at the progression of diversity of bacteria as infections progress. The conventional technique has been bacterial culture, which will only show culturable bacteria, and then only the ones that grow the fastest under the culture conditions. The techniques to look at the progression in detail in vivo have only been developed recently, and are still quite expensive.

    The progression of the infection to a monoculture may be a later stage in the course of an infection. A key transition in an infection is when sufficient quorum sensing agents are expressed to trigger quorum sensing and the expression of virulence factors.

    If you look at their figure 7, suppression of quorum sensing prevented expression of salt-tolerance factors which increased the dose needed for proliferation by in some cases at least 10^5.

    Some bacteria put out compounds that interfere with the quorum sensing of other bacteria. A mixed culture (of the right types) would suppress quorum sensing and even virulent strains would not cause infection. This is how people can be colonized with MRSA and not have an infection. The non-MRSA strains are keeping the MRSA in check. I think that when those inhibiting strains are knocked out by the antibiotics, the pathogen has an open niche and can proliferate. I am pretty sure that is the mechanism by which Pseudomonas became dominant following antibiotic use in the paper I linked to. The strains that were more sensitive to the antibiotics were knocked out first, and their suppression of PA was eliminated which accelerated PA growth and virulence.

    I keep coming back to NO because that is what I know a lot about. During sepsis, the NO level gets very high, perhaps 10 nM/L, high enough to cause generalized vasodilatation, the low blood pressure of septic shock. I think that this may be a “feature”, to try and prevent what ever bacteria are floating around in the blood from attaching somewhere and forming a biofilm. NO does inhibit biofilm formation in SA and in PA (and likely other things too). It also inhibits attachment of malaria infected erythrocytes. I suspect that is why the immune system generates so much NO, to try and prevent biofilm formation.

    If you suppress expression of virulence factors, you prevent infection. Infective strains of the bacteria may be there, but if they don’t express virulence factors they are not virulent and are not causing an infection.

Comments are closed.