JAC Advance Access originally published online on July 24, 2006
Journal of Antimicrobial Chemotherapy 2006 58(3):651-656; doi:10.1093/jac/dkl288
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Antimicrobial practice |
Inappropriate prescribing of aminoglycosides: risk factors and impact of an antibiotic control team
1 Service de Réanimation médicale, Hôpital Henri Mondor (AP-HP) Créteil, France 2 Unité de Contrôle Epidémiologie et Prévention de l'Infection, Hôpital Henri Mondor (AP-HP) Créteil, France 3 CeRBEP-INSERM U657, Institut Pasteur Paris, France 4 Service de Pharmacie, Hôpital Henri Mondor (AP-HP) Créteil, France
*Correspondence address. Unité de Contrôle Epidémiologie et Prévention de l'Infection, Hôpital Henri Mondor, 51 avenue du Mal de Lattre de Tassigny, 94010 Créteil, France. Tel: +33-1-49814691; Fax: +33-1-49814598; E-mail: philippe.lesprit{at}hmn.aphp.fr
Received 25 April 2006; returned 20 May 2006; revised 14 June 2006; accepted 23 June 2006
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionTransparency declarationsReferences |
|---|
Objectives: Many factors are involved in the appropriate use of aminoglycosides, such as modalities of administration, drug levels monitoring and duration of treatment. We assessed prospectively the impact of an antibiotic control team on the appropriateness of prescriptions.
Methods: After a first observational audit assessing the appropriateness of prescriptions, and issuing updated recommendations, we performed an interventional audit, where an antibiotic control team provided counselling when prescriptions were considered inappropriate. Appropriateness of prescriptions, clinical outcomes of patients and medical costs were compared between the two periods.
Results: One hundred consecutive prescriptions were analysed in each period, and 32% of prescriptions were modified by the control team. As compared with the observational period, prescriptions in the intervention period were more appropriate with regard to treatment duration (73% versus 56%, P = 0.01) and drug levels monitoring (61% versus 40%, P = 0.05), and the median treatment duration was shorter (4 versus 6 days, P = 0.0002). Independent factors associated with appropriate treatment duration were hospitalization in intensive care unit [adjusted odds ratio (aOR), 4.46; 95% confidence interval (CI), 1.60–12.46], polymicrobial infection (aOR, 3.97; 1.32–11.92) and antibiotic control team intervention (aOR, 2.49; 1.27–4.87). The intervention period was associated with lower rate of nephrotoxicity (15% versus 4%, P = 0.01) and lower direct medical costs (
4039.4 per 100 persons treated).
Conclusions: Aminoglycoside use was frequently associated with excessive treatment duration and incorrect drug level monitoring. Reinforcing practice guidelines through direct counselling improved the appropriateness of prescriptions.
Keywords: monitoring , prescription , audit
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionTransparency declarationsReferences |
|---|
The aminoglycosides remain frequently used in hospitalized patients for the treatment of endocarditis and severe infections due to Gram-negative bacilli or staphylococci. Many factors are involved in their optimal use, including dosing regimen, drug level monitoring and duration of treatment.1 The above factors are of particular importance, because of the narrow therapeutic index and potential toxicity of these drugs. Because aminoglycosides exhibit both concentration-dependent bactericidal activity and a post-antibiotic effect, a single-daily dosing (SDD) regimen has been advocated to improve drug activity and prevent bacterial resistance.2,3 Besides, SDD regimens may reduce both drug toxicity and the need for monitoring drug levels. However, the SDD concept has been challenged by some experts, and the literature has not resolved all questions about the indications, optimal dosage or the frequency of drug level monitoring during its use.4
For all these reasons, we decided to assess prospectively variables associated with the appropriateness of aminoglycoside prescribing; we also examined the impact of an antibiotic control team in our hospital, a facility where a restricted aminoglycosides dispensation system was operative.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionTransparency declarationsReferences |
|---|
Hospital setting, antibiotic policy and study design
Henri Mondor Hospital is a tertiary-care, 960 bed university hospital, with an average 25 000 admissions per year. Since May 1996, the dispensation of all aminoglycosides has been controlled by introducing a restrictive order form, which provides information on daily dosage and costs of treatment. Orders are checked upon receipt by pharmacists for adherence to indications, selection of the drug and usual daily dosage, as recommended in the hospital's guidelines issued by the Drugs and Therapeutics Committee. Within this context, we repeatedly evaluated five criteria for appropriateness of aminoglycoside prescriptions (justification of prescribing, adequacy of drug choice, adequacy of SDD regimen, indication and timing of drug levels assays, and duration of treatment), first during an unobtrusive observational audit, and second during an interventional audit, after updated guidelines had been issued.
Observational audit
The first audit consisted of a 1 month prospective, unobtrusive observation of 100 consecutive aminoglycoside prescriptions (March 2001). The pharmacist (A. B-C.) notified one of the investigators within 48 h after a new patient had been prescribed an aminoglycoside. All prescriptions delivered for treatment of suspected or documented infections, except for patients in the haematology unit (where specific guidelines are used), were included in the study. Antibiotic prescriptions were grouped into three categories: (i) empirical treatment, including all prescriptions given in the absence of microbiological documentation throughout the antibiotic course; (ii) empirical-secondary documented, which includes all antibiotic courses that were initiated empirically, but were subsequently associated with microbiologically documented infection; (iii) documented infections, with full documentation of infection and identification of a likely pathogen available at the start of treatment.
The investigators were senior physicians with expertise in infectious diseases (J-R. Z., C. R. and P. L.). They recorded on a standardized form the relevant clinical and microbiological data available when treatment was initiated, as well as follow-up information at 3–5 days after the start of treatment and at completion of therapy. Investigators assessed the appropriateness of aminoglycoside use (see below) but did not interfere with physicians' prescribing.
Intervention audit
New recommendations on aminoglycoside use were issued in January 2002, based on the results from the first audit. These included revised recommendations on dosing regimens, and new recommendations on indications and timing of drug levels monitoring. The new recommendations were implemented hospital-wide during early 2002: all staff and junior physicians at our institution were provided with the updated recommendations, and an educational session was held in each ward with the corresponding physicians. The interventional audit occurred a few months later (January 2003), again including prospectively 100 consecutive new prescriptions. It was conducted following the same evaluation protocol as in the first audit, except that the initial evaluation was followed by the intervention of the antibiotic team to provide counselling to prescribing physicians when appropriate.
Evaluation criteria of aminoglycoside use
The investigators determined the appropriateness of aminoglycoside use, based on the following criteria. (i) Justification of use: aminoglycoside prescription was considered justified for patients with neutropenic fever; septic shock; endocarditis; complicated pyelonephritis; post-neurosurgical meningitis; documented infections due to Pseudomonas, Enterococcus or Listeria species; or severe infections due to Staphylococcus or Enterobacteriaceae. (ii) Selection of drug: gentamicin was recommended for treatment of suspected or documented infections due to Gram-positive species and community-acquired Gram-negative bacteria, whereas amikacin was recommended for treatment of nosocomial-acquired infections due to Gram-negative bacteria; tobramycin was restricted to treatment of infections due to Pseudomonas aeruginosa. (iii) Dosing regimen: an SDD regimen was considered adequate in all patients except those with endocarditis, neutropenic fever, specific indications for prolonged therapy of more than 10 days (e.g. endocarditis, bone-joint infection) or severe infections due to staphylococci or P. aeruginosa. (iv) Drug levels monitoring: drug levels (including trough and peak determinations) were to be assayed in patients with renal failure or those treated for more than 5 days. (v) Duration of treatment: this should not exceed 5 days except for endocarditis or severe infections due to antibiotic-resistant Gram-negative bacilli or P. aeruginosa. The appropriateness assessments were reviewed by at least two investigators, and any disagreement in interpretation and rating was resolved by consensus.
Clinical outcome and cost analysis of the intervention
The secondary objectives of the study were to evaluate the impact of the intervention on clinical outcomes and costs. Clinical outcomes assessed included in-hospital mortality and rate of nephrotoxicity. Nephrotoxicity was assessed as a 25% or a 50% increase in the serum creatinine level.2,5
Estimate of the costs of the intervention was based on the costs of aminoglycoside prescriptions + monitoring drug levels + medical intervention by the antibiotic control team. Drug level assays were estimated to cost 27.65 (one peak or trough measurement) or 32.40 (both measurements). An intervention was estimated to take 20 min of work time, corresponding to the average cost for a visit to a specialist physician in the hospital (23). The costs of the intervention were then compared with the corresponding costs of the observational period, which included only the costs of drugs and of drug level assays. All costs were adjusted to 2005 Euros.
Statistical analysis
Median and interquartile range were calculated for continuous variables and absolute and relative frequencies for categorical variables. Univariate comparison of continuous data was performed using the Student's t-test or the Mann–Whitney U-test when appropriate. Categorical variables were compared using a 
2 test or Fisher exact test. All statistical tests were two-tailed. Each potential risk factor for inappropriate duration of treatment was analysed in a univariate model and then all factors with a P value of 0.10 or less were entered in a logistic regression model. Model's goodness-of-fit was assessed using the Hosmer–Lemeshow 2 test and the level of discrimination of the model was measured by the area under ROC curve. The accuracy of the model was defined as good when the area under ROC curve ranged from 0.70 to 0.80 and as excellent above 0.80. Continuous variables were introduced without dichotomization. The adjusted Odds ratio and their 95% confidence interval were calculated for all independent significant predictors. A P value <0.05 was considered significant. Statistical analysis was performed with Intercooled STATA 8.2 software (StatCorp LP, Texas, USA).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionTransparency declarationsReferences |
|---|
Patient characteristics and aminoglycoside prescriptions
The characteristics of the 100 patients included in each period are shown in Table 1. During the intervention period, patients were more frequently admitted to a medical ward and had a lower serum creatinine level on admission than patients included in the observational period. Overall, 34% of the patients were at risk for aminoglycoside-induced nephrotoxicity because of acute or chronic renal failure, including 8% with haemodialysis. Aminoglycosides were most commonly (53.5%) administered for the treatment of hospital-acquired infections. The main sites of infection treated were urinary tract infection (21%), pneumonia (19%) and skin-soft tissue infection (14.5%). Bacteraemia was documented in 27% of the patients.
|
Microbiologically documented infection accounted for most (79%) aminoglycoside prescriptions and included 52% initially documented and 27% empirical/documented infections. The distribution of microorganisms identified did not differ between the observational and intervention periods.
Comparison of the appropriate use of aminoglycosides between the two periods
Of the five appropriateness criteria analysed, two (justification for prescribing and selection of the aminoglycoside) achieved high appropriateness ratings during both periods (Table 2). Two-thirds of prescriptions were associated with adequate administration modalities, but this did not change during the intervention period. Likewise, no significant increase in the rate of prescription of SDD regimen occurred between the two periods. In contrast, we observed during the intervention period a significant improvement in the percentage of guideline-concordant treatment duration, and the median duration was significantly shorter during the intervention period, as compared with the first observational period. The intervention period was also associated with a significantly higher proportion of patients having drug levels monitored adequately: although the overall number of measurements significantly decreased, a higher proportion of patients had peak levels ordered when appropriate, and a higher proportion had drug levels within the therapeutic targets.
|
Among the 100 prescriptions evaluated during the intervention period, 32 were modified by the antibiotic control team. Reasons for intervention included excessive duration of treatment, n = 27 (60%); errors in dosage, n = 6 (13%); lack of indication, n = 5 (11%); inadequate selection of the drug, n = 4 (9%), and requirement for monitoring drug level, n = 3 (7%).
Variables associated with appropriate treatment duration
Several variables were significantly associated with an appropriate duration of treatment in the bivariate analysis (Table 3): intensive care unit location, intervention of the antibiotic team and treatment of a documented infection (either at the start of therapy or secondarily), of a polymicrobial infection, or of pneumonia. On the other hand, age, hospitalization in a surgical ward, hospital-acquired infection, monomicrobial P. aeruginosa and skin-soft tissue infections were significantly associated with inappropriate duration of treatment.
|
All these variables but three (hospitalization in surgery, documented treatment and P. aeruginosa infection, because of collinearity with other variables) were entered into a regression logistic model. Three factors remained independently associated with an appropriate duration in the multivariate analysis, including hospitalization in ICU, polymicrobial infection and intervention of the antibiotic control team (Table 4). Model calibration was good as assessed by the Hosmer–Lemeshow goodness-of-fit test (
2, 5.70; 10 degrees of freedom; P = 0.680). The accuracy of the model was good as ascertained by area under the ROC curve of 0.755.
|
Comparison of clinical outcome and costs of the two periods
There was no difference in hospital mortality of patients treated during the two periods. A significant reduction of the rate of nephrotoxicity was noted during the intervention period, whether assessed as a 25% or a 50% increase in serum creatinine (Table 5). Despite the added costs induced by the medical intervention, overall direct costs associated with aminoglycosides prescriptions were also reduced, because of reduced costs of drugs and of drug levels monitoring. Overall, direct savings were estimated at 4039.4 per 100 persons treated during the intervention period.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionTransparency declarationsReferences |
|---|
Previous studies evaluating antibiotic use in hospitals have shown that up to 50% of prescriptions can be inappropriate.1,6–8 Aminoglycosides are characterized by a narrow therapeutic index and their appropriate use requires careful consideration of many factors; therefore, auditing and monitoring the use of these drugs appears important.9–12
Our study shows that although the SDD regimen had been adopted and its use is generally appropriate according to our guidelines, the administration modalities of aminoglycosides often remain suboptimal both regarding monitoring and duration of therapy. Intervention of an antibiotic therapy team significantly improved the prescribing pattern. A recent audit of gentamicin use also showed that despite provision of guidelines and promotion of SDD regimen in the hospital, aminoglycoside prescribing practices were inadequate with regard to initial dosing and therapeutic drug levels.13
This study was performed in the setting of restricted dispensing of aminoglycosides, implemented by the hospital's Drugs and Therapeutic Committee several years earlier. At the time of the observational audit, orders were checked upon receipt by pharmacists for consistency with existing guidelines, including justification of prescription according to the clinical setting and/or documented infection, selection of the drug and recommended daily dosage. However, no specific recommendation on modalities of aminoglycoside administration, duration of treatment, and indications and frequency of drug level monitoring were included in the hospital's guideline at that time. Consistent with this policy, we found that aminoglycoside use was justified in a large majority of prescriptions and that the drug selected was most often concordant with the guidelines. In contrast, we found unsatisfactory results regarding drug level monitoring and duration of therapy. Indeed, both these indicators of aminoglycoside misuse are difficult to control only through restricted access and controlled delivery. Assessing these aspects of good aminoglycoside prescribing involves auditing patients' charts to provide direct counselling to prescribing physicians; alternatively, computer-generated reminders could be used.1,9–12
Many approaches to improve antibiotic usage in hospitals have been taken, including educational programmes, development of a restrictive hospital formulary and requirement for expert approval before or after prescribing some drugs.14 However, restrictive policies and guidelines have met with limited success. Conversely, several studies have shown a positive impact on the adequacy of antimicrobial prescribing through direct counselling provided to the prescribing physician by an antibiotic therapy team.15–18
In the present study, we evaluated whether auditing aminoglycoside usage and reinforcing practice guidelines through direct counselling could have a positive impact on the appropriateness of prescriptions. Indeed, in a previous study performed in our hospital in 1997, we have shown that an interventional audit markedly increased the appropriate use of ciprofloxacin.17 Using a similar approach, we found that direct intervention by an antibiotic control team resulted in altering one-third of aminoglycoside prescriptions. To assess the impact of this intervention, we used an intent-to-treat analysis to compare the two observation periods and found that the intervention was associated with a significant reduction in the duration of treatment. Importantly, this effect remained significant in the multivariate analysis. Moreover, intervention by the antibiotic control team was associated with a significant improvement of drug level monitoring practices. Whereas drug levels were monitored too often in the control observational period, peak levels were insufficiently monitored. In contrast, the intervention led to a significant reduction of the overall number of assays ordered but an increased rate of peak level assays when appropriate, and a higher proportion of peak and trough levels within the therapeutic ranges.
While the positive effect of an aminoglycoside consultation service on the efficiency of dosing and nephrotoxicity has been described previously, our approach also had a positive impact on the cost and quality of care.1,10,13 Consistent with a reduction in the duration of therapy, the intervention was associated with a lower incidence of nephrotoxicity. The intervention was not associated with increased costs. Contrarily, we found some costs savings, mostly due to the reduction of the number of drug level assays ordered. The benefit reported in the present study is probably underestimated since we did not include the indirect costs of prescriptions, such as work time lost for drug administration and the costs associated with increased nephrotoxicity.
Importantly, others factors associated with an appropriate duration of treatment were found in the study. The rather high appropriate use of aminoglycosides in patients hospitalized in ICU could be explained by the specific antibiotic policy used in this ward, where antibiotic prescriptions are daily reviewed and which is directed by the chair of the anti-infectives drug subcommittee. Polymicrobial infections were also associated with an appropriate duration of therapy, and this effect was mostly observed among infections due to P. aeruginosa. On the other hand, we identified some other factors associated with an excessive duration of treatment, namely monomicrobial P. aeruginosa infections, skin and soft tissue infections and hospitalization in a surgical ward. This suggests that a continuous quality improvement programme of aminoglycosides use in our hospital could be focused on specific high-risk patients or hospital wards.
To conclude, we found that improving the appropriateness of aminoglycoside prescriptions should include both restrictive measures and direct counselling, which requires the intervention of an antibiotic control team. This intervention should focus on potentially amenable factors such as duration of treatment and monitoring modalities, perhaps for specific conditions and patients' subgroups. Alternative options require computer-generated automatic reminders, which may be also effective where computerized prescribing is available.
|
Transparency declarations |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionTransparency declarationsReferences |
|---|
None to declare.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionTransparency declarationsReferences |
|---|
1 Dunagan WC, Woodward RS, Medoff G, et al. (1991) Antibiotic misuse in two clinical situations: positive blood cultures and administration of aminoglycosides. Rev Infect Dis 13:405–12.[Web of Science][Medline]
2 Ali MZ and Goetz MB. (1997) A meta-analysis of the relative efficacy and toxicity of single daily dosing versus multiple daily dosing of aminoglycosides. Clin Infect Dis 24:796–809.[Web of Science][Medline]
3 Bailey TC, Little R, Littenberg B, et al. (1997) A meta-analysis of extended-interval dosing versus multiple daily dosing of aminoglycosides. Clin Infect Dis 24:786–95.[Web of Science][Medline]
4 Brown GH, Bertino JS Jr, Rotschafer JC. (2000) Single daily dosing of aminoglycosides. A community standard? Clin Infect Dis 30:440–1.[CrossRef][Web of Science][Medline]
5
Barza M, Ioannidis JP, Cappelleri JC, et al. (1996) Single or multiple daily doses of aminoglycosides: a meta-analysis. BMJ 312:338–44.
6
Gould IM and Jappy B. (1996) Trends in hospital antibiotic prescribing after introduction of an antibiotic policy. J Antimicrob Chemother 38:895–904.
7
Anglim AM, Klym B, Byers KE, et al. (1997) Effect of a vancomycin restriction policy on ordering practices during an outbreak of vancomycin-resistant Enterococcus faecium. Arch Intern Med 157:1132–6.
8
Working Party of the British Society for Antimicrobial Chemotherapy. (1994) Hospital antibiotic control measures in the UK. J Antimicrobial Chemother 34:21–42.
9 Fletcher CV, Giese RM, Rodman JH. (1986) Pharmacist interventions to improve prescribing of vancomycin and tobramycin. Am J Hosp Pharm 43:2198–201.[Abstract]
10
Saunders NJ, Adams DJ, Lynn WA. (1995) A prospective laboratory-based audit of gentamicin use and therapeutic monitoring. J Antimicrob Chemother 36:729–36.
11 Li SC, Ioannides-Denos LL, Spicer WJ, et al. (1992) Prospective audit of an aminoglycoside consultative service in a general hospital. Med J Aust 157:308–11.[Web of Science][Medline]
12
Shrimpton SB, Milmoe M, Wilson AP, et al. (1993) Audit of prescription and assay of aminoglycosides in a UK teaching hospital. J Antimicrob Chemother 31:599–606.
13 Leong CL, Buising K, Richards M, et al. (2006) Providing guidelines and education is not enough: an audit of gentamicin use at the Royal Melbourne Hospital. Intern Med J 36:37–42.[CrossRef][Web of Science][Medline]
14 Paterson DL. (2006) The role of antimicrobial management programs in optimizing antibiotic prescribing within hospitals. Clin Infect Dis 42:S90–5.
15 Byl B, Clevenbergh P, Jacobs F, et al. (1999) Impact of infectious disease specialists and microbiological data on the appropriateness of antimicrobial therapy for bacteremia. Clin Infect Dis 29:60–6.[Web of Science][Medline]
16 Fowler VG Jr, Sanders LL, Sexton DJ, et al. (1998) Outcome of Staphylococcus aureus bacteremia according to compliance with recommendations of infectious disease specialists; experience with 244 patients. Clin Infect Dis 27:478–86.[Web of Science][Medline]
17
Thuong M, Schortgen F, Zazempa V, et al. (2000) Appropriate use of restricted antimicrobial agents in hospitals: the importance of empirical therapy and assisted re-evaluation. J Antimicrob Chemother 46:501–8.
18 White AC Jr, Atmar RL, Wilson J, et al. (1997) Effects of requiring prior authorization for selected antimicrobials: expenditures, susceptibilities, and clinical outcomes. Clin Infect Dis 25:230–9.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. F. P. Oliveira, C. A. Silva, C. D. Barbieri, G. M. Oliveira, D. M. T. Zanetta, and E. A. Burdmann Prevalence and Risk Factors for Aminoglycoside Nephrotoxicity in Intensive Care Units Antimicrob. Agents Chemother., July 1, 2009; 53(7): 2887 - 2891. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Lesprit, T. Duong, E. Girou, F. Hemery, and C. Brun-Buisson Impact of a computer-generated alert system prompting review of antibiotic use in hospitals J. Antimicrob. Chemother., May 1, 2009; 63(5): 1058 - 1063. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||