Antibiotic Utilization for Acute Respiratory Tract Infections in U.S. Emergency Departments

DISCUSSION

Over the ten-year study period (2001 to 2010), there were more than 12 million annual ED visits for ARTI, with antibiotics used in the majority of these visits. While we observed a decrease in ARTI antibiotic use among patients aged ≤19 years, we observed no decrease in ARTI antibiotic utilization among adult patients, even for those ARTIs where antibiotics are not routinely indicated. Among antibiotic-appropriate ARTI visits, utilization was generally stable, with only three-quarters of patients receiving antibiotics. These results highlight the urgent need to reduce inappropriate use of ARTIs in the ED setting and provide better treatment for those who could benefit from antibiotic therapy.

We provide current estimates of ED antibiotic utilization for ARTI treatment in the United States. To date, most analyses of ARTI antibiotic utilization have focused on outpatient settings, with few examining use in the ED (1, 17, 18). Grijalva and colleagues reported antibiotic prescribing for ARTIs in physician's offices, outpatient clinics, and EDs during 1996 to 2006. While those authors observed overall decreases in antibiotic utilization for patients <50 years old during the period, there were no changes in practice within EDs for all ages combined (1). Neuman and colleagues examined the use of antibiotics for the treatment of pneumonia in the ED during 1993 to 2008, finding an increase in antibiotic use concordant with Infectious Disease Society of America guidelines, as well as an increase in discordant use (19). Our results complement those from the Neuman study and provide updated estimates for ED antibiotic utilization, examining among all age groups a broader range of ARTIs.

Shapiro et al. also recently examined antibiotic use in the ambulatory care setting, reporting a utilization rate of 51% for adult ARTI visits where antibiotics are rarely indicated (20). Importantly, 80% of the antibiotics given for these ARTIs were broad spectrum (20). Our ED-specific results support these estimates, identifying a utilization rate for adult ARTI visits to the ED that was slightly higher than the rate for all ambulatory visits. Our study differed from the work by Shapiro and colleagues in that we examined a longer study period, included all ages, and provide information on trends in antibiotic utilization.

The current report confirms that EDs provide care to an increasingly larger number of patients with ARTIs. This is likely multifactorial and may result from lack of insurance, lack of primary care access, or patient preference to seek care in the ED setting (21). Our results support the hypothesis that many U.S. EDs are functioning as “safety-net” care centers, with the majority of ARTI patients being uninsured or insured by Medicaid (9, 21–23). The observed lack of change in antibiotic utilization for adult ARTI patients, especially those ARTIs for which antibiotics are not indicated, is concerning. This may indicate that efforts to curtail inappropriate antibiotic use have not been effective or have not yet been implemented for this subset of patients. Sustained antibiotic use among adult ARTI patients is likely attributable to a mixture of factors, including patient expectations and the ED environment (24). Specifically, the complexity of ARTI treatment in the ED and the difficulty of making a definitive diagnosis contribute to inappropriate use (25).

Inappropriate use of antibiotics can lead to the development of antibiotic resistance and increase susceptibility to resistant infections (3, 26, 27). However, inappropriate use can also result in substantial morbidity and mortality in a more direct manner, placing individuals at increased risk of antibiotic-related complications. Shehab et al. estimated over 142,000 annual ED visits for complications due to antibiotic use, with nearly 80% related to allergic reactions (4). In addition, antibiotic use and the risk of Clostridium difficile infection have become important concerns (5, 6). Quinolone antibiotics in particular have been shown to cause significant collateral damage and toxicity (i.e., QT prolongation [prolongation of time between the start of the Q wave and the end of the T wave in the heart's electrical cycle], drug interactions, and blood glucose fluctuation) (28, 29). For these reasons, it is important that actions are taken to reduce inappropriate use in the ED and prevent unnecessary morbidity resulting from exposure to antibiotics.

Our findings highlight opportunities for reducing inappropriate antibiotic use among adult ED ARTI patients and for optimizing treatment for antibiotic-appropriate ARTI. Antimicrobial stewardship programs (ASPs) have become a standard practice at U.S. hospitals but have focused primarily on inpatients. Recent literature highlights the success of ASPs in reducing inappropriate antibiotic use in both outpatient and inpatient settings through seminars, roundtable discussions, and personal feedback (25, 30). However, the ED has unique challenges that may not be amenable to standard ASPs. For example, emergency physicians may not be willing to stop and consult antimicrobial guidelines given the high-volume, high-acuity nature of the ED. Doctor-patient relationships in the ED are episodic, and thus ED patients may be less willing to accept emergency physician advice on antibiotic use. Many ED patients do not have access to follow-up primary care, diminishing the options for later adjustment of ARTI care. The optimal approach to antibiotic stewardship in the ED remains unknown but could involve a combination of patient education, rapid diagnostic testing, ED-specific guidelines and treatment pathways, antibiotic order forms, or postprescription reviews (25, 30–32).

We acknowledge several important limitations of the current study. For the study period examined, NHAMCS does not provide the required information to differentiate drugs which were prescribed at discharge from those which were utilized during treatment in the ED, affecting our ability to characterize these distinct patient groups. We also could not assess the duration of treatment or readmission. An additional limitation is that NHAMCS uses a retrospective, probability-sampled design. However, the methodology of NHAMCS is rigorous, and the data set has been widely used in previous antibiotic utilization studies. We were also unable to determine whether ED visits represented readmissions by the same person. Because NHAMCS collects only three diagnoses per patient, we may have missed ARTI visits. Abstractors also may not have been consistent in the selection of diagnoses, resulting in potential misclassification.

An additional limitation is that we were unable to determine granular aspects of individual ED visits that would allow definitive judgment of appropriate antibiotic use. However, by excluding admitted patients and those with suspected bacterial infections, we were able to define a population of ARTI patients whose diagnostic codes suggest that receipt of antibiotics was likely not warranted. Of note, the NHAMCS data set does not contain sufficient information for severity adjustment and does not contain laboratory values or other measures which would allow more conclusive determination of infection severity.

Our definition of ARTI was based on ICD-9 codes, which makes it difficult to truly differentiate certain ARTIs. Due to the nature of ED care, diagnosis of these conditions is often based on nonspecific symptoms and chest radiography. Despite this limitation, NHAMCS abstractors thoroughly review patient charts prior to determining the diagnosis codes included for a given record. Differentiating bronchitis and viral pneumonia from bacterial pneumonia, or viral nasopharyngitis from bacterial rhinosinusitis, can be particularly difficult in the ED setting. In a prior study, positive predictive values of claims-based coding algorithms for pneumonia identification ranged from 72.6% to 80.8%, with sensitivity ranging from 47.8% to 66.2% and specificity ranging from 98.7% to 99.1% (33). Similar estimates were provided for other ARTIs using claims data (34). We feel that low sensitivity would result in conservative estimates of ARTI rates but would not bias our results, as there is no reason to suspect that coding practices would have changed over the study period.

In conclusion, ARTI visits and inappropriate antibiotic use for ARTI remain important problems in the ED, particularly among adult patients. Interventions to reduce inappropriate use of antibiotics which have historically targeted outpatient or inpatient settings must be expanded to the ED setting.