Operational Biosurveillance

Antimicrobial Resistance Forecasting: What Do You Do With The Information?

Physicians with hospital privledges are told to pay attention to the antibiograms produced by the hospital's clinical laboratory.  These are the reports on community resistance to various antibiotics, which is important to note for the clinician because it will often portend success or failure in treatment.  If the patient is sick enough, such decisions may have serious consequences for the patient if the wrong antibiotic is chosen.  

 Below is an example assessment utilizing forecast algorithms for antimicrobial resistance based on a real world local hospital.  Several factors were noted in this assessment:

1. Healthcare provider prescribing behavior that is predicated on experience.  The inherent risks of such prescribing behavior were believed to be greater if the provider was a locum tenens (i.e. a provider moving between hospitals in different states or localities).  The implication was that such providers tended to rely on the Sanford Guide as opposed to local antibiograms.
2. That, as taught in medical school (yet typically ignored by providers), local antibiograms trump general guides such as Sanford.
3. That no medical institution to our knowledge routinely uses forecasts of antimicrobial resistance to guide antimicrobial resistance stewardship program implementation.  It is our experience that forecasts provide additional locale-specific information beyond what the retrospective antibiograms provide.

The dependencies in prescriber behavior follow the points above, such that if prescribers are not using a "local medical intelligence"-informed approach that 1) includes a locality's antibiogram coupled to 2) a locality's forecasted antimicrobial resistance patterns, then antimicrobial stewardship may fall short.  And patient care outcomes may not be optimized.

 

Operational Forecasting of Antimicrobial Resistance

We previously highlighted this topic nearly a year ago.  A crucial topic, as we are seeing a war of attrition with more microbes reported to be capable of side-stepping an increasing array of antibiotics.

About two months ago our hospital isolated its first pan-resistance Pseudomonas strain.  This was a bit of a wake up call for our sleepy community, who had already long gotten over the initial angst of seeing community acquired MRSA.  Odd to say, but few truly get excited about MRSA anymore.  Just another routine medical problem to treat...

To-date, we have maintained 264 pathogen-antibiotic forecast lines and made a few discoveries:

• That agreement between the "old standard", the Sanford Guide, and our forecast library-informed recommendations ranged between 60-100%, depending on the pathogen.  What this implied was physicians not using antibiograms and relying on Sanford to make their decisions ran a tremendous risk of choosing an antibiotic for which the pathogen was resistant.

• That the forecast lines challenged our prior assumptions based on the last antibiogram, where agreement between the a forecast-informed versus last antibiogram-informed recommendation to use a particular drug for a particular pathogen ranged between 67 and 100%.  In other words, if you relied on the last antibiogram, you ran a tremendous risk of continuing to use an antibiotic that may not be available for use within the next 5 years.  Kind of like trying to drive a car forward by looking into the rearview mirror.

• That the largest discrepancies noted between use of the last antibiogram versus being informed by forecast was associated with E. faecium, P. aeroginosa, MSSA, E. faecalis, and E. coli... yet not so for Klebsiella or MRSA, regardless of whether isolated systemically or from urine.  In other words, we were seeing faster apparent changes in antimicrobial resistance patterns for these pathogens.  A bit counterintuitive when we have all been bombarded with concern about MRSA.

As with all forecasts in our domain, we are now challenged with the question of how to exploit this information to maximum benefit for the community.  The answer is not as simple as pulling the given antibiotic off the list of available drugs within the hospital.  You have to contend with:

• "cat herding" among the private commercial pharmacies in the community
• "cat herding" among physicians who are set in their behavioral patterns of prescribing
• differences of engagment between hospital versus independent private outpatient physicians
  
• ongoing resistance to observe proper dispensing of antibiotics in the agricultural, veterinary, and human medical communities
• whether one should instead pick a group of antibiotics and randomly select and rotate antibiotics used for a given type of infection versus picking the one antibiotic favored by the analysis... until that drug becomes unavailable due to evolutionary selection for resistance strains

The problem is a fascinating one as we continue to tease the pieces apart...

 

The Black Canyon Infectious Disease Forecast Station

The first National Weather Service-inspired infectious disease forecast station in US history (aka the Black Canyon Infectious Disease Forecast Station) is approaching its one-year anniversery of operations, now with an online presence and full connectivity to the community of Delta County, Colorado via social media.

 

"Frequency-Modulated Shielding" of a Hospital Using Operational Forecasts For Antibiotic Resistance

One of the greatest challenges we face in medicine is the judicious management of our armamentarium of antibiotics as we confront resistant microbes in our community.  There continues to be debate in the literature regarding the ultimate best strategy for deciding which antibiotic, by pathogen, one should use now versus a year from now when resistance profiles may have migrated. 

The Black Canyon Forecasting Station recently operationalized over 250 pathogen-antibiotic forecast lines for one of our client hospitals.  The major challenge of the hospital is how to manage its antibioitic armamentarium with an eye towards the future.  In other words, "when will we lose the ability to use a given antibiotic for a given pathogen, and based on this knowledge, should we continue to use it?"  Sometimes it's best to holster that "weapon" until a later time when it might be the only option left to you as a physician.

This is a key question from a variety of perspectives- pharmaceutical supply & demand, temporal and financial cost to develop new antibiotics, and patient treatment costs, among many others.  What was interesting to us was the disparity between guidance from The Sanford Guide To Antimicrobial Therapy 2011, Epocrates, and current reality in this particular hospital.  With antibiotic resistance forecasting now a part of the analysis, these disparities become even more poignant.  An example is Pseudomonas aeroginosa, which is horribly resistant to multiple antibiotics, and the forecasts are dismal.

Wiesch et al recently published a nice review of "Population Biological Principles of Drug-Resistance Evolution in Infectious Diseases" (see attached), which reminds us of

The emergence of resistant pathogens in response to selection pressure by drugs and their possible disappearance when drug use is discontinued are evolutionary processes common to many pathogens.

Constant forecasting and real-time monitoring of drug resistance may be our best defense given the research, development, and distribution of new antibiotics is not keeping up with the global problem of resistance. 

For those willing to take a leap from reality, look to the Star Trek-themed "frequency-modulated shielding" of a starship (i.e. hospital).  A bit of silliness, but a highly relevant analogy when considering the very real threat of encroaching drug-resistant bacteria and the reality of a too-slowly evolving antibiotic armamentarium. 

As the Black Canyon Forecasting Station brings this new forecasting capability on-line at the hospital, we note this may be the first time this has ever been attempted...

Download Population biological principles of drug-resistance evolution in infectious diseases

Operational Forecasting of Antimicrobial Resistance

We now have the ability to provide operational forecasts for antimicrobial resistance.  This is going live at the Black Canyon Infectious Disease Forecasting Station soon.