Effect of a Health Care System Respiratory Fluoroquinolone Restriction Program To Alter Utilization and Impact Rates of Clostridium difficile Infection

ABSTRACT

Fluoroquinolones are one of the most commonly prescribed antibiotic classes in the United States despite their association with adverse consequences, including Clostridium difficile infection (CDI). We sought to evaluate the impact of a health care system antimicrobial stewardship-initiated respiratory fluoroquinolone restriction program on utilization, appropriateness of quinolone-based therapy based on institutional guidelines, and CDI rates. After implementation, respiratory fluoroquinolone utilization decreased from a monthly mean and standard deviation (SD) of 41.0 (SD = 4.4) days of therapy (DOT) per 1,000 patient days (PD) preintervention to 21.5 (SD = 6.4) DOT/1,000 PD and 4.8 (SD = 3.6) DOT/1,000 PD posteducation and postrestriction, respectively. Using segmented regression analysis, both education (14.5 DOT/1,000 PD per month decrease; P = 0.023) and restriction (24.5 DOT/1,000 PD per month decrease; P < 0.0001) were associated with decreased utilization. In addition, the CDI rates decreased significantly (P = 0.044) from preintervention using education (3.43 cases/10,000 PD) and restriction (2.2 cases/10,000 PD). Mean monthly CDI cases/10,000 PD decreased from 4.0 (SD = 2.1) preintervention to 2.2 (SD = 1.35) postrestriction. A significant increase in appropriate respiratory fluoroquinolone use occurred postrestriction versus preintervention in patients administered at least one dose (74/130 [57%] versus 74/232 [32%]; P < 0.001), as well as in those receiving two or more doses (47/65 [72%] versus 67/191 [35%]; P < 0.001). A significant reduction in the annual acquisition cost of moxifloxacin, the formulary respiratory fluoroquinolone, was observed postrestriction compared to preintervention within the health care system ($123,882 versus $12,273; P = 0.002). Implementation of a stewardship-initiated respiratory fluoroquinolone restriction program can increase appropriate use while reducing overall utilization, acquisition cost, and CDI rates within a health care system.

INTRODUCTION

Fluoroquinolones are one of the most frequently prescribed antibiotic classes in the United States (1). Based on institutional guidelines, between 30 and 81% of fluoroquinolone use is inappropriately prescribed within the emergency department and inpatient hospital settings (2, 3).

In the United States, the proportion of fluoroquinolone resistance to Escherichia coli and Pseudomonas aeruginosa recovered from patients hospitalized with pneumonia was 30 to 40% between 2009 and 2012 (4). In the treatment of respiratory tract infections, routine utilization or overuse of fluoroquinolones can result in emergence of fluoroquinolone-resistant pneumococci, limiting treatment options (5). Adverse consequences, such as Clostridium difficile infection (CDI), have been associated with fluoroquinolone use (6–9). In addition, there is speculation that the incidence of CDI may be higher with antianaerobic fluoroquinolones, such as moxifloxacin, due to a potential for greater disruption in normal flora. A reduction in fluoroquinolone use may also lead to a reduction in CDI (10–12). Antimicrobial stewardship strategies have been shown to reduce fluoroquinolone utilization by 30 to 70% (13–17). Fluoroquinolone restriction can also lead to reduced C. difficile infections, but large-scale studies from the health care system perspective, particularly focused on respiratory fluoroquinolones, are lacking. The objective of the present study was to assess the impact of a respiratory fluoroquinolone restriction program on fluoroquinolone utilization, appropriateness of quinolone-based therapy, and CDI rates.

(This study was presented in part at IDWeek 2014, Philadelphia, PA, October 2014, and at ASM Microbe/ICAAC 2016, Boston, MA, June 2016.)

RESULTS

A decrease in both respiratory fluoroquinolone utilization and CDI rates was experienced in both the education and restriction phases compared to the preintervention phase (Fig. 1). Compared to preintervention, the four hospitals experienced 48 and 88% average reductions in utilization (DOT/1,000 PD) after education and restriction, respectively. Respiratory fluoroquinolone utilization decreased from a monthly mean and standard deviation (SD) of 41.0 (SD = 4.4) DOT/1,000 PD preintervention to 21.5 (SD = 6.4) DOT/1,000 PD and 4.8 (SD = 3.6) DOT/1,000 PD posteducation and postrestriction, respectively. Using segmented regression analysis, both education (14.5 DOT/1,000 PD per month decrease; P = 0.023) and restriction (24.5 DOT/1,000 PD month decrease; P < 0.0001) were associated with decreased respiratory fluoroquinolone use in the four hospitals without a prior respiratory fluoroquinolone stewardship program (Fig. 2). In addition, mean and SD monthly CDI cases/10,000 PD decreased by roughly 50% from 4.0 (SD = 2.1) preintervention to 2.2 (SD = 1.35) postrestriction. CDI rates decreased significantly (P = 0.044) from preintervention using education (3.43 cases/10,000 PD) and restriction (2.2 cases/10,000 PD) (Fig. 3). Compared to baseline rates (44.4 DOT/1,000 PD), the average rate of other commonly used intravenous antibiotics did not change significantly during the posteducation (45.2 DOT/1,000 PD) or postrestriction time period (47.2 DOT/1,000 PD).

FIG 1.

Respiratory fluoroquinolone utilization and CDI rate throughout the study time period. Utilization is expressed as days of therapy per 1,000 patient days (DOT/1,000 PD) on the left x axis, and the CDI rate is expressed as the number of C. difficile infection cases per 10,000 patient days (CDI/10,000 PD) on the right x axis.

FIG 2.

Respiratory fluoroquinolone utilization expressed as days of therapy per 1,000 patient days (DOT/1,000 PD) in intervention and control hospitals during the study time period.

FIG 3.

C. difficile infection rates (CDI/10,000 PD) in study hospitals during the study time period.

There was no levofloxacin use throughout the study period. A total of 481 individual orders for moxifloxacin were reviewed for appropriateness: 232 in the preintervention group and 249 in the postrestriction group (Table 1). All patients in the preintervention group received at least one dose of moxifloxacin versus 130/249 (52%) in the postrestriction group (P < 0.001). In the postrestriction group, an intervention was performed on 119 orders which resulted in discontinuation of moxifloxacin prior to administration of the first dose, yielding 130 patients who ultimately received moxifloxacin. In addition, 191/232 (82%) and 65/249 (26%) of patients received two or more doses in the preintervention and postrestriction groups, respectively (P < 0.001). A significant increase in appropriate respiratory fluoroquinolone use was experienced postrestriction versus preintervention in patients given at least one dose (74/130 [57%] versus 74/232 [32%]; P < 0.001), as well as those receiving two or more doses (47/65 [72%] versus 67/191 [35%]; P < 0.001). In the postrestriction group, 38/65 (58%) patients administered only one dose of moxifloxacin did not meet criteria for utilization. Of these, pharmacists intervened on 26/38 (68%) of patients prior to administration of a second dose. Of note, there were significantly more orders for chronic obstructive pulmonary disease (COPD)/bronchitis/sinusitis in the preintervention group (P < 0.01) and significantly more orders for endophthalmitis/globe rupture (P = 0.03), surgical prophylaxis (P < 0.01), and infections due to mycobacteria (P = 0.02) in the postrestriction group (Table 1). However, in the postrestriction group, approximately 55% (6/11) of patients with an order for moxifloxacin with an indication of surgical prophylaxis were switched to an alternative antibiotic prior to receiving a dose. The assessment of beta-lactam allergies also contributed to the reduction of moxifloxacin utilization. Although there were more reported beta-lactam allergies in patients with moxifloxacin orders in the postrestriction versus preintervention group (97/249 versus 61/232, P = 0.003), 38% fewer patients with reported allergies were administered moxifloxacin (60/97 versus 61/61, P < 0.001). No adverse events were reported during the data collection period.

TABLE 1.

Moxifloxacin orders and consistency with restriction criteria

Parameter	No. (%) of orders		P
	Preintervention (n = 232)	Postrestriction (n = 249)
No. of doses administered
≥1	232 (100)	130 (52)	<0.001
1	41 (18)	65 (26)	0.03
≥2	191 (82)	65 (26)	<0.001
Consistency: no. of doses administered according to restriction criteria
≥1 dose administered	74 (32)	74 (57)	<0.001
1 dose administered	7 (17)	27 (42)	<0.01
≥2 doses administered	67 (35)	47 (72)	<0.001
Ordered indication
Community-acquired pneumonia	121	146	0.15
COPD/bronchitis/sinusitis	69	37	<0.01
Other	21	14	0.15
Urinary tract infection	9	9	0.88
Skin and soft tissue infection	6	5	0.77
Intra-abdominal infection	3	4	1.00
Endophthalmitis/globe rupture	3	12	0.03
Surgical prophylaxis	0	11	<0.01
Mandibular osteomyelitis	0	4	0.12
Mycobacterial infection	0	7	0.02

Despite no change in purchase price per unit, a significant reduction in the annual moxifloxacin acquisition cost was found postrestriction compared to preintervention within the health care system ($123,882 versus $12,273; P = 0.002). No differences in Streptococcus pneumoniae susceptibilities for moxifloxacin, ceftriaxone, or penicillin were demonstrated during the study period within the four hospitals (P = not significant).

DISCUSSION

Decreasing antimicrobial selection pressure to promote antibiotic resistance is an established goal of antimicrobial stewardship programs (18). Respiratory fluoroquinolones have been shown to be overused for indications in which other antimicrobials are more appropriate or for which antibiotics are not indicated, such as in the case of treatment of acute exacerbations of chronic bronchitis (19, 20). This is especially important given the increased resistance with fluoroquinolones and a strong association with onset of CDI (6–9). Strengths of the present study include a large cohort of hospitals from within a health care system, a comparator hospital in which the policy was already established, and a separate assessment of education and restriction stewardship policies. This study provides a unique method for health care systems to decrease the inappropriate use of respiratory fluoroquinolones.

Formulary restriction is a recommended strategy for reducing antimicrobial utilization within the guidelines for developing an institutional antimicrobial stewardship program (18). A previous single center study found significantly less moxifloxacin use before (1,038 ± 109 defined daily doses [DDD]/month) compared to after (42 ± 10 DDD/month) implementation of a formulary restriction policy in a community teaching hospital; however, this study also found that partial utilization was replaced with levofloxacin (11). Our results confirm and expand upon this single-center study through restriction of all respiratory fluoroquinolones and also present reductions in days of therapy (DOT) per 1,000 patient days, the reporting metric recommended by the guidelines for implementing an antibiotic stewardship program (21).

The assessment of beta-lactam allergy to avoid unnecessary fluoroquinolone use is also unique and our finding that these assessments impacted patients who received two or more doses (due to the delay in assessment) will require further study. Indications for respiratory fluoroquinolone use were also deemed to be more appropriate using hospital-standardized guidelines for use. There were significantly fewer orders for COPD/bronchitis/sinusitis (deemed inappropriate indications) and more orders for endophthalmitis/globe rupture and infections due to mycobacteria (deemed appropriate indications) in the postrestriction group. These findings likely reflect practice change over time as prescribers became accustomed to the restriction criteria. Although studies have questioned the effectiveness of education, our study did find a significant reduction in utilization of moxifloxacin after education (roughly 50%), suggesting that targeted education with opportunity for provider feedback may be an effective strategy within a health care system.

Prior studies have shown increased risk of CDI after receipt of fluoroquinolones (6–9); however, data on antimicrobial stewardship strategies restricting respiratory fluoroquinolone use and impact on CDI are scarce particularly in the endemic setting. Two prior studies have demonstrated significant reduction of CDI after fluoroquinolone restriction. Single-center studies have shown decreased CDI rates postimplementation of moxifloxacin education and restriction ([59 ± 3] versus [32 ± 3] CDI cases/month; P = 0.0044) (11). A two-center, community hospital study assessed the impact of fluoroquinolone (levofloxacin and ciprofloxacin) restriction (phase 1) and removal from ward stock (phase 2) on CDI with similar results (12). Our present study provides additional confirmation on a health care system level demonstrating that a decrease in moxifloxacin use can lead to a reduction in CDI rate.

Limitations to the present study include the nonrandomized nature of the respiratory fluoroquinolone restriction. However, we controlled for this by ensuring that no additional CDI-related interventions, including those related to infection prevention, occurred during the time period and by also including a control hospital in which the intervention had already been established. The control hospital allowed for assessment of any potential fluctuations in respiratory fluoroquinolone utilization that would have led to reduction due to other variables. In addition, a reduction in fluoroquinolone use was not associated with an increase in the use of other commonly used intravenous antibiotics. However, the intervention was not blinded, and it is possible that other variables may have influenced the decreased CDI rates observed.

Data on the appropriateness of respiratory fluoroquinolone therapy was collected retrospectively; however, variables commonly present in the electronic health record were chosen to assess appropriateness, and data collection occurred in the same manner in the “pre” and “post” time periods. In addition, although all respiratory fluoroquinolones were targeted during this initiative, moxifloxacin was the formulary respiratory fluoroquinolone, with levofloxacin remaining nonformulary. Whether a similar intervention in health care systems that have different formulary respiratory fluoroquinolones (i.e., levofloxacin) would have the same effect, particularly on CDI rates, is unknown. Antianaerobic fluoroquinolones such as moxifloxacin may be associated with a higher incidence of CDI due to a potential for greater disruption in normal flora. Although all fluoroquinolones have been associated with CDI (6–9), conflicting data exist regarding the potential for the increased risk of CDI with antianaerobic fluoroquinolones. Three studies have reported increased CDI rates after formulary substitution from levofloxacin to an antianaerobic fluoroquinolone (i.e., moxifloxacin or gatifloxacin) (22–24); however, only two of the studies detected decreased rates after a formulary change back to levofloxacin (23, 24). In addition, Jump et al. performed a retrospective analysis of reported CDI rates within health care facilities located in Ohio and found similar rates of CDI when the data were grouped by formulary status for levofloxacin (8.5 cases/10,000 PD, 95% confidence interval [CI] = 7.83 to 9.3) or moxifloxacin (8.5 cases/10,000 PD, 95% CI = 7.8 to 9.2), respectively (25).

Lastly, we did not perform strain typing of our isolates, which may limit study generalizability. This is of particular interest for ribotype 027, which often displays greater resistance against fluoroquinolones compared to the nonepidemic strain (26); therefore, enhanced reductions in CDI rates may be more prominent in hospitals with a higher prevalence of this particular ribotype.

In conclusion, we found that an antimicrobial stewardship-initiated respiratory fluoroquinolone program, including education and restriction, improved appropriate respiratory fluoroquinolone use while reducing overall utilization, as well as CDI rates.

MATERIALS AND METHODS

This was a multicenter, quasiexperimental study conducted at four of twelve adult hospitals within Seton Healthcare Family, a large, urban, not-for-profit health care system located throughout Central Texas. The four hospitals ranged from 124 to 534 licensed beds. The health care system utilized a dual-fluoroquinolone formulary consisting of ciprofloxacin and moxifloxacin. Prior to the study, one hospital within the health care system had already implemented an antimicrobial stewardship program focused on respiratory fluoroquinolone reduction through prospective audit by an infectious diseases pharmacist. The pilot project was associated with approximately 50% decrease in CDI rate at the one hospital; thus, the Antimicrobial Stewardship Committee devised a strategy to expand this program system-wide, which included the development of respiratory fluoroquinolone restriction criteria, as well as a two-phase intervention, including education and implementation of restriction criteria for utilization. This provided a unique opportunity to study the effect of the intervention throughout four other hospitals within the health care system using the original hospital as a comparator. The collection of data as part of a research study in conjunction with this clinical initiative was approved by the Seton Healthcare Family Institutional Review Board.

Development of the respiratory fluoroquinolone policy.

Prior to this study, an extensive literature review was performed to guide the initial development of institutional treatment guidelines, including community-acquired pneumonia and antibiotic therapy in COPD exacerbations. These literature findings and expert opinion were used to develop education material, respiratory fluoroquinolone restriction criteria (Fig. 4), and institutional treatment guidelines. The restriction criteria were developed by the health care system's Antimicrobial Stewardship Committee in collaboration with the critical care medical directors at the four medical centers. Moxifloxacin remained the health care system's formulary respiratory fluoroquinolone. These criteria were approved by the health care system Pharmacy and Therapeutics Committee. To avoid overuse of respiratory fluoroquinolones in patients with a suspected beta-lactam allergy, a beta-lactam allergy assessment tool was also developed that guided pharmacists on appropriate steps for allergy evaluation.

FIG 4.

Respiratory fluoroquinolone restriction criteria for utilization.

Clinician education on appropriate use of respiratory fluoroquinolones was conducted over a 3-month period (September to November 2013) led by two infectious disease pharmacists. Education included presentations and emails to key stakeholders focused on the literature justification of institutional treatment guidelines and utilization criteria for respiratory fluoroquinolones. Providers targeted for education included critical care and emergency medicine physicians, hospitalists, surgeons, internal medicine and family medicine physicians, and residents. Physicians were encouraged to provide feedback regarding the proposed criteria for utilization. All pharmacists within the health care system were required to complete a competency-based electronic training module regarding the criteria for utilization and proper beta-lactam allergy assessment. The respiratory fluoroquinolone restriction policy was initiated December 2013. The policy included pharmacist prospective review of all respiratory fluoroquinolone orders for consistency with the approved restriction criteria, a beta-lactam allergy assessment on patients with orders for respiratory fluoroquinolones and a reported allergy, and direct contact with prescribers when patients did not meet criteria for utilization to recommend alternative therapy.

Evaluation of policy effectiveness.

Respiratory fluoroquinolone utilization data and hospital-acquired CDI rates were collected monthly preintervention (April 2013 to August 2013) and postintervention stratified by posteducation time period (September to November 2013) and postrestriction time period (December 2013 to December 2014) for the four medical centers. In addition, utilization data were collected throughout the study time period for other commonly utilized intravenous antibiotics (ampicillin-sulbactam, aztreonam, cefazolin, cefepime, cefoxitin, ceftriaxone, ciprofloxacin, clindamycin, meropenem, and piperacillin-tazobactam) to assess any changes that may impact CDI. The acquisition cost for moxifloxacin, the health system's formulary respiratory fluoroquinolone, was assessed for the health care system during the same study periods. Hospital-acquired CDI was identified within the four hospitals by trained infection control practitioners using the National Healthcare Safety Network (NHSN) definition (27). PCR was utilized for detecting C. difficile. Finally, S. pneumoniae susceptibilities were reviewed for the four hospitals based on annual antibiograms (July 2012 to June 2015) to assess susceptibilities during the study time frame. To assess the appropriateness of respiratory fluoroquinolone use after implementation of the restriction criteria, orders within the four study hospitals were evaluated in the first 6 months of the postrestriction time period (January to June 2014) and compared to a similar number of randomly selected patients within the preintervention time period. Data collection included antibiotic indications, duration of therapy, reported beta-lactam allergies and reaction, prior tolerance of beta-lactams, and adverse events.

Statistical analysis.

Respiratory fluoroquinolone orders, indications, and appropriate utilization based on restriction criteria were analyzed in the preintervention and postrestriction phases using a chi-square or the Fisher exact test. Utilization data (DOT/1,000 PD) were assessed using segmented regression analysis, with the hospital with an already established fluoroquinolone restriction policy serving as a control. The average days of therapy for other commonly used intravenous antibiotics were compared using segmented regression analysis. CDI rates (CDI/10,000 PD) at baseline, during the education period, and during the postrestriction period were compared using the Kruskal-Wallis test. Moxifloxacin annual expenditure data were analyzed between calendar years 2013 and 2014 according to the Student t test. S. pneumoniae susceptibilities for moxifloxacin, ceftriaxone, and penicillin were assessed using one-way analysis of variance to assess the variance between the study time periods.