Wagging the Dog: Antibiotic Use and the Emergence of Resistance

In recent years, we have witnessed an increase in antimicrobial resistance that threatens to render our antibiotic arsenal useless. If current trends continue, the real possibility exists that we will return to a period not unlike the preantibiotic era when we helplessly watched as infectious diseases exacted a tremendous toll worldwide. Antibiotic resistance increases infection-related morbidity and mortality and contributes greatly to the cost of health care both through prolonged hospital stay and the need for more expensive drugs.^1,2 Increasing antimicrobial resistance is of great concern not only because the agents available to treat these resistant infections are limited, but also because, for better or worse, the treatment of infectious diseases relies heavily on empiric therapy. Indeed, it is probable that this spiraling empiricism has contributed significantly to the emergence of resistance. Subsequently, empiricism has become increasingly difficult as the susceptibility patterns of common pathogens become more and more complicated and unpredictable.

Antimicrobial resistance has been traditionally viewed as a nosocomial problem. Frequent reports of hospital-acquired pathogens such as methicillin-resistant Staphylococcus aureus(MRSA), vancomycin-resistant enterococcus(VRE), and extended-spectrum β–lactamase (ESBL)-producing gram-negative organisms support this perception. However, recent years have witnessed the emergence of resistance in community-acquired pathogens such as Haemophilus influenzae, Streptococcus pneumoniae³Campylobacter jejuni.⁴ Similarly, antimicrobial resistance has also emerged in the organisms that most frequently cause urinary tract infections (UTIs). A recent study noted that the percentage of urinary pathogens resistant to ampicillin increased from 29% to 38% from 1992 to 1996, while resistance to trimethoprim-sulfamethoxazole (TMP-SXT) increased from 8% to 16% during the same time period.⁵ Eighteen percent of Escherichia coli, the organism responsible for the vast majority of UTIs, were resistant to TMP-SXT.⁵

The treatment of UTIs is most often empiric. Increasing resistance to TMP-SXT, a common first line empiric agent, has complicated this strategy. In this issue of the Journal of General Internal Medicine, Wright and colleagues report the results of a study designed to identify factors associated with TMP-SXT resistance in outpatient urinary coliform isolates.⁶ This retrospective case-control study was conducted in an Emergency Department (ED) and compared patients with resistant and susceptible isolates. Multivariable analysis revealed diabetes, recent hospitalization, current use of antibiotics, and recent use of TMP-SXT to be risk factors for TMP-SXT resistance. When patients with a recent hospitalization were excluded from analysis, only current use of antibiotics and recent use of TMP-SXT remained independent predictors of resistance.⁶

Although the findings of this study strongly suggest that prior antibiotic use predisposes to subsequent resistant infection, several limitations regarding the patient population should be addressed. First, the presence of infection was never definitively documented; although symptoms associated with UTI were sought in the medical record, no urinalysis data was collected. Therefore, some of the isolates may represent urinary tract colonization rather than true infection. This would be important if risk factors for the two conditions differ. In addition, it is unclear whether all patients had urine cultures obtained because a UTI was suspected or if some cultures were obtained for other reasons (i.e., routine admission workup, other genitourinary tract complaints). Similarly, the authors did not report the indications for prior therapy among patients who were receiving antibiotics at the time of their ED visit. One might postulate that many of these patients were receiving empiric therapy for a UTI and presented to the ED because of treatment failure. If this were true, the association of prior antibiotic use and resistant infection would have been overestimated. In addition, recall bias should be considered as it is possible that patients with a resistant pathogen may be more likely to recall recent antibiotic use, or be more thoroughly questioned about recent use by a physician. Finally, the generalizability of the study must be questioned inasmuch as patients who have urine cultures obtained in an ED likely differ from those who have them obtained in a physician's office or those who are treated empirically. One could postulate that patients who visit the ED may have a greater acuity of illness and have less access to medical care.

Despite these concerns, the authors have demonstrated an association between antibiotic use, particularly the use of TMP-SXT, and TMP-SXT resistance in urinary coliform pathogens. A recent matched case-control study conducted in a pediatric Canadian population showed that prior antibiotic use, genito-urinary tract abnormalities, and prior hospital admission were all independent risk factors for TMP-SXT resistance in E. coli urinary isolates.⁷ The findings of these studies suggest that risk factors may be similar in adult and pediatric populations. Although there is little prior data linking antibiotic use and resistant UTIs, the results of Wright and colleagues are not surprising. In fact, the association of antecedent use of antibiotics and antimicrobial resistance has been well-established for many years,⁸ but causality has yet to be definitively established.

The implications of rising TMP-SXT resistance for treatment of urinary tract infections are less certain. Indeed, it is not at all clear whether resistance results in poorer outcomes in this setting. Wright et al. did not compare the clinical or microbiological outcomes of patients with resistant and susceptible isolates.⁶ Interestingly, it has been noted that the bacteriological cure rate of urinary tract infections is not affected by whether the organism is susceptible or resistant to a given agent.⁹ This may be because the levels of antibiotic achieved in the bladder are higher than those used in susceptibility testing or perhaps because a subinhibitory concentration of antibiotic combines with the natural bladder defenses to eradicate the infecting organism. For these reasons, empiric therapy with TMP-SXT may continue to be appropriate despite increasing resistance. Alternatively, as alluded to by the authors, one could develop a clinical prediction rule that would allow more accurate classification of patients as high or low risk for resistant infection. In patients determined to be at high risk of resistant infection, empiric therapy could be modified, particularly in those patients in whom waiting for culture results would be undesirable (e.g., those who are more severely ill or who are unlikely to follow-up).

The results of this study might lead some to consider routine empiric fluoroquinolone therapy in order to provide more consistent activity against the urinary coliform bacteria; the authors actually make this recommendation for certain populations. We believe this would be at best a short-term solution and should not be the conclusion drawn from this study. The authors themselves provide the historical perspective that TMP-SXT replaced ampicillin as the empiric agent of choice for UTIs because of the emergence of resistance. It is equally likely that now favoring empiric therapy with a fluoroquinolone would only repeat this cycle. Sequential changes in antibiotic use will only serve to “squeeze the balloon;” this is a concept proposed by Burke,¹⁰ which contends that substitution policies instituted in response to emerging resistance will only result in new resistance patterns. Indeed, it has been noted that the prevalence of resistance to fluoroquinolone antibiotics has increased as the use of these agents has become more widespread, even among more intrinsically susceptible Enterobacteriaceae such as E. coli.¹¹ Furthermore, prior receipt of a fluoroquinolone antibiotic has been noted to be an independent risk factor for fluoroquinolone-resistant E. coli UTIs.¹² The ramifications of increasing fluoroquinolone resistance would be grave given the many clinical situations in which these agents are currently used.

With respect to this issue, it appears as though the tail is wagging the dog instead of the dog wagging its tail. Progressive changes to broader spectrum antibiotics in response to emerging resistance will ultimately deplete our antimicrobial armamentarium. The only rational approach is to focus on research designed to understand and improve physician antibiotic prescribing practices. Wright et al. do not address the indications for antibiotic use in patients who presented on therapy; it is likely that many were treated unnecessarily. Focus groups of doctors have suggested that up to 50% of outpatient antibiotics are unnecessary.¹³ Upper respiratory infections and bronchitis account for nearly one-third of outpatient antibiotic prescriptions,¹⁴ despite the fact that these conditions are almost always viral in origin and antimicrobial agents have shown little clinical benefit in these infections.¹⁵ Comprehensive educational interventions have been successful in decreasing antibiotic use for these conditions and were not associated with worse clinical outcomes.¹⁶ Furthermore, limitations on antibiotic use have been shown to be effective in reversing antimicrobial resistance patterns. For example, a Finnish study noted that the proportion of group A streptococci resistant to macrolide antibiotics decreased by almost 50% following efforts to limit the use of these antibiotics.¹⁷

The emergence of antimicrobial resistance is increasing at an alarming rate in both community and hospital-based settings. Changes in recommendations for empiric therapy will likely promote the emergence of further resistance and subsequently exhaust the agents available to treat certain infections. Therefore, we must focus on efforts to define and understand the relationship between antimicrobial use and the emergence of resistance, and should develop and study programs to improve the use of antibiotics. Only through this understanding and the judicious use of antibiotics can we hope to slow the ever-widening scope of antimicrobial resistance in an effort to once again regain our advantage over these pathogens.—Ebbing Lautenbach, MD, MPH,Neil O. Fishman, MD,University of Pennsylvania Medical Center, Philadelphia, PA.