Article, Urology

Impact of an antimicrobial stewardship intervention on urinary tract infection treatment in the ED

a b s t r a c t

Study objective: The study objective is to assess changes in treatment of uncomplicated Urinary tract infections after implementation of recommendations based on national guidelines and local resistance rates.

Methods: This preintervention and postintervention study included patients discharged home from the emergen- cy department (ED) with an uncomplicated UTI at a 439-bed teaching hospital. Emergency department prescribers were educated on how local Antimicrobial resistance rates impact UTI practice guidelines. Empiric treatment according to recommendations was assessed as the primary outcome. Agreement between chosen therapy and isolated pathogen susceptibility was compared before and after education. Reevaluation in the ED or hospital admission within 30 days for a UTI was also evaluated.

Results: A total of 350 patients were studied (174 before and 176 after education). Of those, 255 had cystitis, and 95 had pyelonephritis. After education, choice of therapy consistent with recommendations increased from 44.8% to 83% (difference, 38.2%; 95% confidence interval [CI], 33%-43%; P b .001). The change was predominately driven by an increase in nitrofurantoin use for cystitis from 12% to 80% (difference, 68%; 95% CI, 62%-73%; P b .001). Agreement between empiric treatment and the isolated pathogen susceptibility improved for cystitis 74% to 89% (P = .05), and no change occurred in 30-day repeat ED visits for a UTI.

Conclusions: After implementation of treatment recommendations for uncomplicated UTIs based on local resistance, empiric antibiotic selection improved in the ED. To further meet goals of antimicrobial stewardship, additional interventions are needed.

Introduction

Background

The Infectious Diseases Society of America (IDSA) and the European Society for Microbiology and Infectious Diseases published updated practice guidelines for uncomplicated cystitis and pyelonephritis in women during 2011 [1]. Because of the large variance in Escherichia coli resistance to fluoroquinolones and trimethoprim-sulfamethoxazole (TMP-SMX) throughout the world, the guidelines place a large emphasis

? Author affiliation at the time of study was HSHS St John’s Hospital.

?? No financial support to disclose.

? Meetings: Interscience Conference of antimicrobial agents and Chemotherapy 2014,

Washington, DC.

* Corresponding author at: Drake University College of Pharmacy and Health Sciences, 2507 University Ave, Des Moines, Iowa 50311. Tel.: +1 515 271 2110.

E-mail addresses: [email protected], [email protected]

(K.M. Percival), [email protected] (K.M. Valenti), [email protected] (S.E. Schmittling), [email protected] (B.D. Strader), [email protected]

(R.R. Lopez), [email protected] (S.J. Bergman).

on the importance of using local resistance rates to determine the best empiric treatment [1,2]. Specifically, the guidelines recommend that TMP-SMX no longer be used as first-line therapy for uncomplicated cystitis when local resistance for E. coli exceeds 20%. If patients are being discharged on oral therapy for pyelonephritis in areas where

E. coli resistance to fluoroquinolones exceeds 10%, then it is also suggested that a 1-time dose of a long-acting parenteral agent from a different antimicrobial class be used. Another addition to the guidelines is the concept of avoiding collateral damage, which includes the eco- logical adverse effects from antibiotic therapy, especially the selec- tion of multidrug resistant organisms such as methicillin-resistant Staphylococcus aureus, vancomycin-resistant Enterococci, and Clostridium Difficile[1,3]. It emphasized that drugs with minimal impact on the micro- biota, such as nitrofurantoin, should be used when possible, whereas higher risk drugs, including fluoroquinolones, should be reserved for infections more severe than cystitis [1].

The public health threat of antimicrobial resistance and the need to prevent its spread are at the forefront of importance as demonstrated by the White House releasing an Executive Order and National Strategy to Combat Antibiotic-Resistant Bacteria in September 2014 [4]. The

http://dx.doi.org/10.1016/j.ajem.2015.04.067

need for increased education on antimicrobial resistance and selection of therapy was demonstrated in a recent survey of health care providers that revealed antibiotic resistance was not commonly considered when prescribing antimicrobials despite the widespread concern for resis- tance [5]. The definition of antimicrobial stewardship according to guidelines from IDSA and the Society of Healthcare Epidemiology of America is “to optimize clinical outcomes while minimizing unintended consequences of antimicrobial use, including toxicity, the selection of pathogenic organisms (such as C. difficile), and the emergence of resis- tance” [6]. antimicrobial stewardship programs have successfully dem- onstrated the ability to safely reduce resistance by emphasizing use of narrow-spectrum antimicrobials, but these efforts have largely focused on inpatient settings despite most prescribing for antibiotics occurring in outpatients. A call to action has been expressed for antimicrobial stewardship in the emergency department (ED), as this clinical setting can impact antibiotic use in both inpatients and outpatients [7]. Litera- ture regarding antimicrobial stewardship in the ED has been minimal, possibly due to the difficulties of implementation in this setting. Some of these challenges include rapid patient turnover, the diverse needs of those destined for either inpatient or outpatient care, a varied mix of providers, and high staff turnover. A few of the interprofessional antimicrobial stewardship processes that have led to positive patient outcomes in the ED include development of ED-specific antibiograms as well as postprescription culture follow-up [7-9]. The educational intervention in this study was designed to assist prescribers in making the best choice of therapy for uncomplicated UTIs, the infectious disease requiring the most frequent culture review.

Goals of this investigation

The purpose of this study was to assess changes in treatment of uncomplicated UTIs after implementation of recommendations derived from applying local antimicrobial resistance patterns to national practice guidelines. It is hypothesized that increasing adherence to the guidelines will result in a higher rate of isolated pathogens being susceptible to the prescribed therapy.

Methods

Study design and setting

This was a quasi-experimental study comparing 2 separate periods before and after an educational intervention in November 2013. It was conducted at a 439-bed tertiary-care teaching center with more than 57, 000 ED visits annually and approximately 100 visits per month for the diagnosis of any UTI. The facility has had an antimicrobial steward- ship service since 2011, but no previous attempts have been made to implement outpatient interventions. clinical pharmacists are present in the ED 24 hours each day. They verify patient Medication orders during their stay but do not evaluate outpatient prescriptions unless requested. A postprescription review program is in place to follow up with patients who have positive culture results after discharge. Outcome measures were compared for the months after November 2012 and November 2013. The institutional review board approved the study before beginning research and provided a waiver of the requirement for informed consent, as the intervention was deemed to be of minimal risk to subjects.

Selection of participants

Patients evaluated for a UTI by any ED provider during the specified periods were eligible for the study. This included examinations by resident physicians, midlevel practitioners, and attending physicians. Patients were identified through ED visit reports by diagnosis and were included if they were female, aged 12 to 70 years old, discharged home from the ED with an uncomplicated UTI, and received an

antibiotic prescription. Exclusion criteria included patients who were admitted for inpatient treatment, pregnant, catheterized, or diagnosed with a complicated UTI for any other reason.

Interventions

With the help of the microbiology laboratory, an ED-specific antibiogram was constructed to determine the rate of E. coli resistance to TMP-SMX, ciprofloxacin, nitrofurantoin, and cefazolin locally. For outpatients, the antibiogram revealed susceptibilities for these drugs of 75%, 80%, 99%, and 96%, respectively. Based on this and the IDSA guide- lines, institution-specific recommendations were developed for the em- piric treatment of uncomplicated UTIs in the ED as shown in Table 1.

After being endorsed by our local antimicrobial stewardship com- mittee, institution-specific recommendations were implemented through education by a pharmacist in the ED and to resident physicians during their monthly meeting. In addition, all ED providers were deliv- ered education by e-mail from the medical director of the ED reinforcing the recommendations and their justification. Pharmacists did not ac- tively review outpatient prescriptions during the study. A preliminary audit of empiric prescribing was performed 2 months into the posteducation study period, and then feedback on results was provided by e-mail to all ED providers as a reminder of the recommendations. The remaining patient charts were reviewed at the end of the study period.

Methods and measurements

All data were extracted systematically from the electronic medical record by 1 trained investigator using a standardized data collection form with definitions of each variable. Diagnosis of UTI was based on provider documentation and The International Classification of Diseases, 9th Revision, Clinical Modification codes assigned to the visit. If the type of UTI was not specified, classification was based on evidence-based definitions [10]. Patients were determined to have cys- titis if there was no documentation of flank pain, fever, leukocytosis (white blood cell >= 12 000/mL), or the prescriber recorded that there was no evidence for pyelonephritis. An infection was defined as uncom- plicated if it occurred in nonpregnant woman with no known urologic abnormalities. If documentation was not clear, the abstractor reviewed the available data with the senior investigator. If there was a discrepan- cy, then a third investigator was consulted for interpretation. The study

Table 1

Empiric treatment recommendations for acute uncomplicated UTIs

Cystitis

- - First choices
    - Nitrofurantoin monohydrate/macrocrystals 100 mg every 12 h for 5 d
      - Only for patients with CrCl N 60 mL/min

-or-

- - - Cephalexin 500 mg every 12 hours for 7 days
  - Second choice
    - Trimethoprim-sulfamethoxazole 160/800 mg every 12 h for 3 d
      - Appropriate in patients with CrCl 15-60 mL/min and a ?-lactam allergy
  - Third choice
    - Ciprofloxacin 250 mg every 12 h or levofloxacin 250 mg daily for 3 d
      - Appropriate in patients with CrCl b 60 mL/min plus a contraindication to

?-lactam and sulfa Pyelonephritis

- - A urine culture and susceptibility should always be performed
  - Give 1 dose of a long-acting parenteral agent in the ED
    - Ceftriaxone 1 g, gentamicin or tobramycin 5 mg/kg (pharmacy to dose)
  - Oral prescription for discharge
    - First choice
Ciprofloxacin 500 mg every 12 h for 7 d or levofloxacin 750 mg daily for 5 d
- - - Second choice
Trimethoprim-sulfamethoxazole 160/800 mg every 12 h for 14 d
- - - Third choice
Cephalexin 500 mg every 6 h for 14 d

Note: Doses are for patients with normal renal function. Adjustment of therapy may be required for patients with kidney disease.

team met regularly to review progress. The drug, dose, frequency, duration, and use of 1-time parenteral injection, if warranted, were assessed according to the recommendations in Table 1. After nitrofurantoin and cephalexin, the use of TMP-SMX was considered appropriate in patients’ with a creatinine clearance (CrCl) 15 to 60 mL/min and a ?-lactam allergy. The fluoroquinolones, ciprofloxacin, and levofloxacin were appropriate in those with CrCl less than 60 mL/ min and a contraindication to both a ?-lactam and sulfa drug. These fluoroquinolones, TMP-SMX, or cephalosporins were considered appro- priate therapy in pyelonephritis in that order. Isolated pathogen suscep- tibilities were compared with Empiric therapy, and reevaluation for a UTI in the ED or hospital admission within 30 days was assessed to determine treatment failure.

Outcomes

The primary outcome of this study was to assess Adherence to recommendations for the treatment of uncomplicated UTIs based on local resistance rates. Secondary outcomes included the agreement between Empiric antibiotics prescribed and isolated pathogen susceptibilities and reevaluation in the ED or hospital admission for a UTI within 30 days.

Analysis

Primary and secondary outcomes were analyzed statistically according to data type. Nominal data were assessed with ?² tests using GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, CA www.Graphpad.com). Continuous data were analyzed with a Student t test performed using Microsoft Excel 2010 (Microsoft, Redmond, WA). For the primary outcome, a sample size of 343 patients was determined necessary to detect a 15% difference with an 80% power. A P value of .05 was considered statistically significant.

Results

Characteristics of study subjects

The flowchart of patients evaluated in the study is shown in the Figure. The most common diagnosis code was for “nonspecific UTI.” Upon review, 255 patients were classified as having cystitis and 95 with pyelonephritis. There were no meaningful differences in the demographics of patients before and after education. Baseline charac- teristics are shown in Table 2. E. coli was the most common pathogen in positive urine cultures for both preeducation and posteducation patients with cystitis (73% and 71%, P = .75) and pyelonephritis (75%

and 58%, P = .23), respectively.

Main results

Antibiotics prescribed at discharge changed significantly after edu- cation (Table 3). Before the intervention, the choice of empiric therapy was consistent with recommendations 44.8% of the time compared with 83% after (difference, 38.2%; 95% confidence interval [CI], 33%- 43%; P b .001). This change was driven by significant decreases in TMP-SMX, and fluoroquinolone use for cystitis balanced with increases in prescribing of nitrofurantoin for cystitis and fluoroquinolones for py- elonephritis. Overall, prescribing according to institution-specific rec- ommendations for the treatment of UTIs in regard to antibiotic choice, dose, frequency, duration, and a 1-time parenteral antibiotic dose for pyelonephritis preeducation and posteducation increased from 2.3% to 20% (difference, 17.7%; 95% CI, 14%-22%; P b .001) (Table 4). The lowest rate of adherence to recommendations was in duration of therapy, which changed from 16% to 25.5% (difference, 9.5%; 95% CI, 6%-13%; P = .029), and administration of a long-acting parenteral agent for pyelonephritis different from the treatment at discharge. There was no further change in prescribing observed after feedback was delivered via e-mail to providers after the initial audit of empiric treatment halfway through the prospective study period.

Patients excluded with complicated UTI

n = 125

Pre-education: n = 53

Post-education: n = 72

Patients discharged from ED with UTI n = 475

Pre-education

n = 174

Patients included n = 350

Pyelonephritis n = 68

Cystitis n = 106

Post-education

n = 176

Pyelonephritis n = 27

Cystitis n = 149

Figure. Study flowchart.

Table 2 Patient characteristics
	Cystitis			Pyelonephritis
	Pre-education	Post-education	P	Pre-education	Post-education	P
Characteristic	n = 106	n = 149		n = 68	n = 27
Mean age, y	31.8 (12.5)	29.5 (11.5)	.14	34.4 (13.9)	32.1 (13)	.47
Mean WBC, k/mL	9.5 (4.5)	8.9 (2.7)	.44	10.7 (5.2)	12.3 (3.7)	.19
Mean temperature, ?C	36.7 (0.4)	36.8 (0.4)	0.48	36.9 (0.6)	36.9 (0.5)	.84
Mean SCr	0.80 (0.2)	0.84 (0.1)	0.15	0.83 (0.2)	0.80 (0.1)	.57
Mean CrCl	88.5 (12.7)	98 (26.2)	0.04	87.2 (17.3)	99.3 (25.7)	.03
Urine culture performed (%)	58 (54.7)	103 (69.1)	.02^a	54 (79.4)	22 (81.5)	.82^a
Mean days of treatment	7 (2.8)	7.2 (2.2)	.46	7.9 (2.9)	8.4 (2.8)	.43

Abbreviation: WBC, white blood cell; SCr, serum creatinine; CrCl, creatinine clearance. All values are number (+-SD) unless specified. All P values by Student t test unless noted.

^a ?² test.

When a urine culture was performed, the prescribed antibiotic was susceptible to the isolated pathogen more often in cystitis after educa- tion (74% vs 89%, P = .05) but not in pyelonephritis patients (90% vs 76%, P = .23). The rate of patients seeking follow-up care for a UTI at the institution within 30 days was unchanged at 4.6% compared with 7.4% (P = .27).

Discussion

We observed that the Prescribing habits for treatment of uncompli- cated UTIs changed to use narrower-spectrum antibiotics after imple- mentation of antimicrobial stewardship recommendations in the ED. This is noteworthy because national guidelines were tied to local resis- tance rates, and providers adjusted empiric prescribing accordingly after education. Because hospital pharmacists do not normally view the prescriptions patients are being discharged home on and communi- ty pharmacists do not have access to the medical record, no other inter- vention was performed during the study period unless a provider asked for assistance. Our results demonstrated the largest differences before and after education in treatment for patients with cystitis, which is a very common diagnosis in ED patients being discharged home. Al- though we were able to show significant improvements in appropriate antibiotic choice, the results indicate further work can be done to opti- mize treatment.

Recently, outcomes have been published from another center repli- cating our improvement in guideline adherence for the treatment of un- complicated UTIs, in their case through the utilization of an electronic order set. That intervention resulted in a 38% increase in adherence to guidelines, primarily from a reduction similar to ours in use of fluoroquinolones for cystitis. In that study, unnecessary antibiotic days were decreased from 250 to 52 per 200 patients [11]. Although our ed- ucation improved the days of therapy prescribed to be more consistent with guidelines, recommendations for treatment duration were only followed a minority of the time. This may have been because our educa- tion focused primarily on attributing rising resistance rates to empiric antibiotic selection as opposed to prolonged duration of therapy. This

leads us to believe that future studies intending to improve antibiotic use should also incorporate utilization of order sets, custom-built with recommended agents based on local resistance and durations of therapy. At our institution, order sets have been difficult to implement during a time of transition between paper and electronic prescriptions, so treatment recommendations were distributed through verbal education with paper handouts provided and e-mail. Although order sets were not included in this study, they will be considered for antimi- crobial stewardship efforts in the future based on the success of this baseline study.

One of the alarming findings from our experience was that patients labeled as having pyelonephritis were being prescribed nitrofurantoin at discharge both before and after education. This is concerning because nitrofurantoin does not achieve adequate concentrations in the kidney tissue and is not appropriate to treat a potentially Systemic Infection. Al- though the retrospective nature of this analysis could have misclassified the infection, it is a point that necessitates further education to pro- viders for the sake of patient safety. We have continued the practice of educating ED providers on UTI treatment recommendations upon the annual arrival of new medical residents.

The most commonly described antimicrobial stewardship interven- tion in the ED is postprescription culture review, and there have previ- ously been improvements in readmission rates for patients with this follow-up. One way pharmacists have assisted with these review pro- grams is to ensure that patients with positive test results are being treat- ed appropriately after they have left the ED [9,12,13]. Our intervention was intended to improve prescribing before the patient was discharged from the ED, subsequently leading to less follow-up that would be need- ed later. In our experience, there was a clinically relevant improvement in the number of times the isolated pathogen was susceptible to the pre- scribed antibiotic. We did not specifically evaluate the number of mi- nutes spent on postprescription review in this study but felt there was a meaningful decrease in the amount of follow-up needed for uncompli- cated UTIs after the intervention because the therapy chosen was active against the isolated pathogen more often. This is especially important in an era of emphasis on cost-effective health care and limited

Table 3

Antibiotics prescribed at discharge

Cystitis Pyelonephritis

Pre-education Post-education P Pre-education Post-education P

n = 106 (%) n = 149 (%) n = 68 (%) n = 27 (%)

Cephalexin 1 (0.9) 2 (1.3) .77 0 1 (18.5) .11

FQs 35 (33) 16 (12.8) b.001 32 (47) 19 (70.4) .04

TMP-SMX 56 (52.8) 8 (5.4) b.001 29 (43) 2 (7.4) .001

Nitrofurantoin 13 (12.3) 119 (79.9) b.001 7(10) 5 (18.5) .28

Doxycycline 1 (0.9) 0 .23 – –

Abbreviation: FQs, fluoroquinolones: ciprofloxacin or levofloxacin. All P values by ?² test.

Table 4

Results of adherence to recommendations

Pre-education, n= 174 (%)

Post-education n = 176 (%)

P value

possibility that patients initially evaluated in the ED could have sought follow-up care for their UTI at another site. This would lower our estima- tion of treatment failures both before and after the intervention. Further- more, the study was powered to show a difference in treatment of 15%

Combined cystitis and pyelonephritis overall 4 (2.3) 35 (20) b.001 Cystitis n = 106 (%) n = 149 (%)

Discharge antibiotic	17 (16)	124 (83.2)	b.001
Dose	87 (82)	138 (92.6)	.02
Frequency	105 (99)	146 (97.9)	.64
Duration	22 (20.8)	35 (23.2)	.65
Overall adherence Pyelonephritis	3 (2.8) n = 68 (%)	32 (21.5) n = 27 (%)	b.001
Parenteral antibiotic in ED	6 (8.8)	3 (11.1)	.73
Discharge antibiotic	61 (89.7)	22 (81.5)	.28
Dose	55 (80.9)	22 (81.5)	.95
Frequency	62 (91.2)	22 (81.5)	.18
Duration	6 (8.8)	10 (37)	b.001
Overall adherence	1 (1.5)	3 (11)	.03

All P values by ?² test.

reimbursement for treatment failures. Our findings did not show a dif- ference in the number of reevaluations within 30 days between the 2 groups. This is most likely due to the fact that even in the preeducation group, culture follow-up was being performed, and patients were contacted immediately if the organism was resistant to empiric treat- ment. They were then changed to appropriate treatment based on the urine culture and susceptibility report. In the meantime, even cystitis with organisms resistant to the prescribed therapy can sometimes be al- leviated due to the high concentrations of most antibiotics in the urine. This would potentially limit the return of patients, although it is not a re- liable way to practice.

Limitations

This was a single-center observational experience without randomiza- tion. Recommendations were based on resistance rates that will differ in other geographic areas, which limit external validity. Our study spanned 4 to 5 months, 1-year apart, and this is not long enough to determine whether cumulative resistance rates would change over time based on the improved adherence to guidelines. Only uncomplicated UTIs were an- alyzed for this study, and the number of patients with a diagnosis of pyelo- nephritis was small, especially in the posteducation group, further limiting the applicability of results. The low number of pyelonephritis cases in both groups may have been due to the definition chosen, as few patients had a fever recorded. Most coding was for “nonspecific UTIs,” and, therefore, classification was based on chart documentation. Despite patients being enrolled prospectively in the second half of the study, all the charts were analyzed retrospectively, and the authors had to apply definitions for py- elonephritis and cystitis themselves. The abstractor was not blinded to study group because of the nature of a before and after trial design. This leaves open the possibility that some patients may have been misclassified, although every effort was made to accurately assess these patients without bias. In addition, the report generated to identify patients evaluated for UTIs in the ED changed slightly from the pre-education to post-education group. The pre-education report was able to identify all The International Classification of Diseases, 9th Revision codes during the visit, whereas the post-education report only identified patients by the primary diagno- sis. This could have contributed to the discrepancy in the number of patients with pyelonephritis between the groups, although it seems unlikely that many patients would be discharged with pyelonephritis as a secondary diagnosis in our experience. Finally, there remains the

but not reevaluation rates of 5%, so a larger trial would need to be conduct- ed to detect if any difference existed in that outcome.

Conclusion

In summary, outpatient prescribing in the ED changed significantly after implementation of treatment recommendations for uncomplicat- ed UTIs based on local resistance patterns and national practice guide- lines. This resulted in achieving a goal of antimicrobial stewardship by decreasing use of broad spectrum agents for cystitis, specifically fluoroquinolones, and reserving them for more severe infections. Subse- quently, there was an increase in isolated pathogens being susceptible to empiric therapy for cystitis after education. Additional studies of an- timicrobial stewardship in the ED are needed to determine the impact interventions have on long-term resistance patterns, time required for post-prescription follow-up, and patient outcomes. In the future, we advise implementing order sets focusing on recommendED treatments including durations of therapy for UTIs and appropriate use of 1-time doses for long- acting parenteral agents before discharge with pyelonephritis.

Acknowledgments

The authors acknowledge the assistance of McKenzie Ferguson, PharmD in study design and statistical analysis.

References

AJEM