Editorials

At the crossroads of the quest for quality and data overload, a priori rational examination of what is to become standard of practice at large expense to our already stretched healthcare system, one must humbly concede that some important events still cannot be counted with a high degree of accuracy. The study by Brown et al (6) at the very least helps clarify the other half of the chiasmus, or what it is that should not be counted, even though it can.

REFERENCES 1. Quote Investigator: Not Everything That Counts Can Be Counted. Available at: http://quoteinvestigator.com/2010/05/26/everythingcounts-einstein/comment-page-1/. Accessed April 3, 2014

2. Cameron WB: Tell me not in mournful numbers. NEA J 1958; 47:173. National Education Association of the United States, Washington, DC. As cited in Quote Investigator. Not Everything That Counts Can Be Counted. Available at: http://quoteinvestigator.com/2010/05/26/ everything-counts-einstein/comment-page-1/. Accessed April 3, 2014 3. Institute of Medicine: To Err Is Human: Building a Safer Health System. Washington, DC, National Academies Press, 1999 4. National Quality Forum: “Find Measures” page. “Critical Care” as query term limited to the “No longer NQF-endorsed” category. Available at: http://www.qualityforum.org. Accessed April 2, 2014 5. Magill SS, Klompas M, Balk R, et al: Developing a new, national approach to surveillance for ventilator-associated events. Crit Care Wed 2013; 41:2467-2475 6. Brown SES, Ratcliffe SJ, Halpern SD: An Empirical Comparison of Key Statistical Attributes Among Potential ICU Quality Indicators. Crit Care Wed 2014; 42:1821-1831

Antimicrobial Utilization Decision Support in the Critical Care Unit: Is the Glass Half-Empty or Half-Full?* Mark E. Rupp, MD Trevor C. Van Schooneveld, MD Department of Internal Medicine University of Nebraska Medical Center Omaha, NE

ecently, the Centers for Disease Control and Prevention released an alarming report noting that more than two million Americans are sickened yearly by antibioticresistant pathogens resulting in at least 23,000 deaths (1). Unfortunately, antibiotic use is the key driver of resistance, and the more antibiotics are used, the more antibiotic resistance ensues (2). In critically ill persons, it is imperative to quickly identify those patients with infection and promptly initiate appropriate anti-infectives. Numerous studies have demon­ strated that inappropriate initial antimicrobial therapy (largely due to antibiotic resistance) is associated with increased mor­ tality and poor outcome (3-5). Herein lies the rub—in order to effectively treat critically ill patients with infections, clinicians are increasingly forced to use combinations of antibiotics with broad-spectrum activity, which further fuels the emergence of

R

‘ See also p. 1832. Key Words: antibiotic resistance; antimicrobial utilization; decision support Dr. Van Schooneveld consulted for Cubist. His institution received grant support from Cubist and ViroPharma. Dr. Rupp has disclosed that he does not have any potential conflicts of interest. Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000388

Critical Care Medicine

resistance and subsequent use of more broad-spectrum antibi­ otics. The solution to this spiraling problem is to quickly and accurately identify patients with infections due to resistant pathogens, promptly initiate appropriate antibiotics, avoid unnecessary antibiotic exposure in uninfected persons, and stop antibiotics as soon as an infection has been adequately treated. Unfortunately, each of these steps is complex and dif­ ficult to achieve. Two broad strategies exist to improve empiric antimicrobial use: 1) rapid identification and determination of the suscepti­ bility of the pathogen and 2) development of prediction tools to identify those at greatest risk of infection due to resistant pathogens. Although great strides have been made in clinical microbiology to speed the identification of pathogens through a variety of techniques (polymerase chain reaction, matrixassisted laser desorption ionization-time of flight, etc.), these methods require technical expertise and capital invest­ ment and are not universally available. Another approach to assist clinicians in making better therapeutic decisions is to use traditional epidemiologic techniques, such as prediction rules, to identify patients who are at greater risk of infection due to resistant pathogens. A drawback of prediction rules is they require the analysis of large amounts of data to produce widely applicable results. The introduction of electronic health records holds promise as they allow easy access to vast amounts of clinical data and the potential to generate real-time clinical alerts. Previous studies have explored the promise of electronic records and clinical alerts. In one of the seminal studies on this topic, Evans et al (6) described the introduction of a com­ puterized system which combined microbiologic, clinical, and w w w .c c m jo u r n a l.o r g

1947

Editorials

pharmacy data to produce individualized antimicrobial rec­ ommendations at the point of care. Compared with historic data, the introduction of this system not only improved the accuracy of empiric therapy (97.8% vs 81.9%, p < 0.01) but also decreased inappropriate prescribing and overall antibiotic use (6). Thursky et al (7) developed a similar computerized system in a single-center critical care unit and noted a decline in inappropriate empiric therapy (24.4% vs 15.2%, p = 0.02) and a 10.5% decrease in antibiotic use. Both of these systems used unit antibiograms and infectious disease expertise to develop recommendations. The problem with such systems is that they do not address individual patient risk factors such as previous exposure to a specific antimicrobials or recent recov­ ery of a resistant pathogen. To help address this difficult problem, Micek et al (8) devel­ oped and retrospectively tested a point-of-care antibiotic use decision support system supported by an electronic medical record. As described in this issue of Critical Care Medicine, criti­ cally ill patients receiving cefepime, piperacillin-tazobactam, or meropenem for healthcare-associated infections due presum­ ably to Gram-negative pathogens were targeted. An electronic alert was generated if a patient had received the same antibiotic class that was being prescribed in the previous 6 months or had a pathogen isolated in the prior 6 months that was resistant to the currently prescribed antibiotic. Alerts were generated on 900 of 3,616 patients; 569 patients (15.7%) had a Gram-nega­ tive pathogen identified and 143 (4%) had a pathogen resistant to the prescribed antibiotic. Alerts were generated on 179 of 569 patients with infection due to Gram-negative organisms and patients triggering an alert were significantly more likely to have received an inappropriate antibiotic (35.8% vs 20.3%). Of those who received inappropriate therapy, 39 of 64 patients could have received an alternative effective antibiotic based on the information in the alert. The authors concluded that their simple automated alert system designed around analysis of only two information items could have identified about 40% of critically ill patients who received inappropriate antibiotic therapy. Thus, is the glass 45% full or 55% empty? Although it is cer­ tainly worthwhile to utilize a fairly simple and economical tool to identify nearly half of patients receiving inappropriate anti­ biotics, the system has a number of limitations. Nine hundred alerts were generated among 3,616 patients to identify only 64 of 143 patients (44.8%) receiving inappropriate therapy. This means 14 alerts were generated for each patient ultimately found to be receiving inappropriate therapy (positive predic­ tive value of 7.1%). Clinical decision support systems have been previously noted to generate many nonactionable alerts potentially leading to “alert fatigue” (9). As the authors noted, the performance characteristics of the alert system are far from

1948

w w w .c c m jo u r n a l.o r g

perfect: 44.8% sensitivity, 75.9% specificity, and area under the receiver operating curve, 0.603. However, as also is noted by the authors, the system could potentially be improved by incor­ porating other easily derived information from the electronic medical record, such as recent hospitalizations, immunocom­ promised status, and residence in a long-term care facility. Also, this was a retrospective analysis in a single center and whether observations are generalizable is unknown. One potentially significant issue in using this system in other centers is that it is dependent on the limited presence of carbapenemases. As these enzymes inactivate all (3-lactam antibiotics, simply shift­ ing patients to a previously unused (3-lactam antibiotic will not be useful. In conclusion, this simple alert system, which is not depen­ dent on sophisticated data mining or expensive technology, could be readily modified and prove clinically useful. When combined with other antimicrobial stewardship efforts in the critical care unit, such as use of biomarkers to identify infec­ tion (e.g., procalcitonin), rapid diagnostics (gene amplifica­ tion), and de-escalation regimens, we will be able to more effectively treat patients while limiting unnecessary antimicro­ bial exposure.

REFERENCES 1. U.S. Department of Health and Human Services: Centers for Disease Control and Prevention. 2013. Antibiotic resistance threats in the United States, 2013. Available at: http://www.cdc.gov/drugresistance/threat-report-2013/index.html. Accessed February 27, 2014 2. World Health Organization: Department of Communicable Disease Surveillance and Response. W H O global strategy for containment of antimicrobial resistance. 2001. Available at: http://www.who.int/ csr/resources/publications/drugresist/EGIobal_Strat.pdf. Accessed February 27, 2014 3. Kang Cl, Kim SH, Kim HB, et al: Pseudomonas aeruginosa bacte­ remia: Risk factors for mortality and influence of delayed receipt of effective antimicrobial therapy on clinical outcome. Clin Infect Dis 2003; 37:745-751 4. Shorr AF, Micek ST, Welch EC, et al: Inappropriate antibiotic therapy in Gram-negative sepsis increases hospital length of stay. Crit Care Med 2011; 3 9 :4 6 -5 1 5. Garnacho-Montero J, Garcia-Garmendia JL, Barrero-Almodovar A, et al: Impact of adequate empirical antibiotic therapy on the outcome of patients admitted to the intensive care unit with sepsis. Crit Care Med 2003; 3 1 :2 7 4 2 -2 7 51 6. Evans RS, Pestotnik SL, Classen DC, et al: A computer-assisted management program for antibiotics and other antiinfective agents. N Engl J Med 1998; 3 3 8 :2 3 2 -2 3 8 7. Thursky KA, Buising KL, Bak N, et al: Reduction of broad-spectrum antibiotic use with computerized decision support in an intensive care unit. Int J QuaI Health Care 2006; 18:224-231 8. Micek ST, Heard KM, Gowan M, et al: Identifying Critically III Patients at Risk for Inappropriate Antibiotic Therapy: A Pilot Study of a Point-ofCare Decision Support Alert. Crit Care Med 2014; 4 2 :1 8 3 2 -1 8 3 8 9. Hermsen ED, VanSchooneveld TC, Sayles H, et al: Implementation of a clinical decision support system for antimicrobial stewardship. Infect Control Hosp Epidemiol 201 2; 33:41 2-4 1 5

August 2014 • Volume 42 • Number 8

Copyright of Critical Care Medicine is the property of Lippincott Williams & Wilkins and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Antimicrobial utilization decision support in the critical care unit: is the glass half-empty or half-full?*.

Antimicrobial utilization decision support in the critical care unit: is the glass half-empty or half-full?*. - PDF Download Free
2MB Sizes 0 Downloads 4 Views