ORIGINAL ARTICLE

Retrospective Study of Prolonged Versus Intermittent Infusion Piperacillin- and in Intensive Care Unit Patients at an Academic Medical Center

Rebekka J. Dow, PharmD,* Warren E. Rose, PharmD,Þ Barry C. Fox, MD,þ Joshua M. Thorpe, PhD, MPH,§ and Jeffrey T. Fish, PharmD||

existing antimicrobials. The application of pharmacodynamic A Background: Prolonged infusions of -lactams are increasingly being principles can aid in the preservation of efficacy, utilized because of increased antimicrobial resistance and a lack of new impede the emergence of resistance, and potentially provide in the pipeline. a pharmacoeconomic benefit.1 A-Lactams are time-dependent Methods: A retrospective, observational study evaluated adult patients killing antibiotics and demonstrate maximum efficacy when the who received at least 72 hours of piperacillin-tazobactam or meropenem percentage of time during which the free drug concentration in a medical and surgical intensive care unit (ICU) at an academic exceeds the minimum inhibitory concentration (%f T 9 MIC) is medical center. In the first 6 months of the study, all patients received optimized. In general, a bactericidal effect occurs when the free conventional intermittent dosing; in the second 6 months, all patients drug concentration exceeds the MIC for 40%, 50%, and 60% received these antibiotics by prolonged infusions. The main outcome to 70% of the dosing interval for the , , comparisons between groups were duration of ventilator support, dura- and , respectively.2,3 Piperacillin-tazobactam and tion of ICU and hospital length of stay, and in-hospital mortality. meropenem both exhibit time-dependent killing and are often Results: A total of 121 patients were included: 54 patients in the inter- used in the treatment of multidrug-resistant bacterial infec- mittent (67% piperacillin-tazobactam, 33% meropenem) and 67 patients tions.1,2,4 Prolonged infusions of these A-lactams increase the in the prolonged group (81% piperacillin-tazobactam, 19% meropenem). time of antimicrobial exposure above the MIC and should as a The prolonged group demonstrated a significant decrease in ventilator result improve their efficacy.1,2,4Y6 days (j7.2 days; 95% confidence interval [CI], j12.4 to j2.4), ICU Population pharmacokinetic modeling and Monte Carlo length of stay (j4.5 days; 95% CI, j8.3 to j1.4), and hospital length of simulations have been used to generate new dosing strategies, stay (j8.5 days; 95% CI, j18.7 to j1.2) compared with the intermittent such as the prolonged infusion of A-lactam antibiotics, and group. The risk of in-hospital mortality was 12.4% in the prolonged in- have been applied directly into clinical practice.4 Piperacillin- fusion group and 20.7% in the intermittent infusion group corresponding tazobactam as a prolonged infusion of 3.375 g in Monte Carlo Y to an odds ratio of 0.54 (0.18 1.66). simulations demonstrates a greater probability of attaining 50% Conclusions: The results of this study show that the use of pro- f T 9 MIC in isolates of reduced susceptibility and has corre- longed infusions of piperacillin-tazobactam and meropenem could po- sponded to improved outcomes in critically ill patients.2,4 tentially improve clinical outcomes in critically ill populations. As The pharmacodynamic profile of meropenem has been antibiotic resistance continues to increase in gram-negative pathogens, evaluated as a prolonged infusion administration.6 Although less A these -lactam dosing strategies will be important for appropriate outcome data in the literature are available for meropenem use treatment. with this dosing strategy, prolonged infusions of meropenem Key Words: piperacillin-tazobactam, meropenem, intensive care unit, have greater %f T 9 MIC and potentially reduced morbidity and 7 A-lactams mortality compared with standard infusions. Y Based on the above evidence, the University of Wisconsin (Infect Dis Clin Pract 2011;00: 00 00) Hospital and Clinics instituted a guideline that was approved by the Pharmacy and Therapeutics Committee for the prolonged ecause of increased antimicrobial resistance and a shortage infusions of both piperacillin-tazobactam and meropenem for B of novel antimicrobial development, new dosing strategies patients receiving this therapy in a medical and surgical intensive have been proposed to optimize the pharmacodynamics of care unit (ICU), on August 1, 2008. This study was conducted after the guideline was instituted to further evaluate the efficacy of prolonged infusion regimens. From the *Department of Pharmacy, Froedtert & The Medical College of The objective of our study was to compare the clinical and Wisconsin; †Pharmacy Practice Division, University of Wisconsin School pharmacoeconomic outcomes of conventional intermittent dos- of Pharmacy; ‡Department of MedicineYInfectious Diseases, University of Wisconsin Hospital and Clinics; §Social and Administrative Sciences ing of piperacillin-tazobactam and meropenem to the prolonged Division, University of Wisconsin School of Pharmacy; and ||Department infusions in critically ill patients, with a proactive focus on re- of Pharmacy, University of Wisconsin Hospital and Clinics, Madison, WI. ducing ventilator days in ventilated patients, length of stay in Correspondence to: Jeffrey T. Fish, PharmD, University of Wisconsin both the ICU and hospital, and patient mortality. Hospital and Clinics, Pharmacy Services, 600 Highland Ave, F6/133-1530, Madison, WI 53792. E-mail: [email protected]. This study has no sources of support. MATERIALS AND METHODS Presented in part at the Great Lakes Pharmacy Residency Conference, West Lafayette, IN, April 29YMay 1, 2009. This study was approved by the University of Wisconsin Dr Fox was previously on the speaker’s bureaus of (which makes Hospital and Clinics Health Sciences institutional review board. piperacillin-tazobactam) and Cubist (which markets meropenem). Data were collected by conducting a retrospective chart review Dr Rose has received research funding and is on the speaker’s bureau for Cubist. All other authors have no competing interests to disclose. of all patients 18 to 89 years of age admitted to the medical and Copyright * 2011 by Lippincott Williams & Wilkins surgical ICU who received at least 72 hours of therapy with ISSN: 1056-9103 piperacillin-tazobactam or meropenem from February 1, 2008,

Infectious Diseases in Clinical Practice & Volume 00, Number 00, Month 2011 www.infectdis.com 1

Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Dow et al Infectious Diseases in Clinical Practice & Volume 00, Number 00, Month 2011 to January 31, 2009. Cystic fibrosis patients, those receiving mortality, patient ventilator days, duration of therapy with continuous renal replacement therapy, or those who were re- piperacillin-tazobactam or meropenem, and other concomitant ceiving therapy in the ICU from July 31 to August 1 and had antibiotics and their duration of therapy. Data on all cultures their regimen changed per the new protocol were excluded. The during the time frame of piperacillin-tazobactam or meropenem first dose of each antibiotic is infused over 30 minutes to rapidly treatment (3 days before treatment initiation to 7 days after obtain therapeutic concentrations. Each subsequent dose is treatment discontinuation), organism susceptibility to the study given per prolonged infusion while the patient is in the ICU. agent used, and site of infection (determined from positive cul- According to the guideline, a patient’s regimen may change ture or clinical assessment) were collected. The defined daily when transferred out of this ICU. For patients with a pathogen dose (DDD) per 1000 patient days in this ICU was collected with an MIC of 2 Hg/mL or greater, the same dosage, interval, before and after implementation of prolonged infusion for both and infusion rate continued on the intermediate care or general study antimicrobials. The APACHE II score was defined as the care unit for both piperacillin-tazobactam and meropenem. For worst physiological score calculated during the initial 24 hours those pathogens with MICs of less than 2 Hg/mL, or if there is after admission to the ICU. Susceptibility testing was performed no recovered organism, the piperacillin-tazobactam or mer- by in vitro Kirby-Bauer testing with zone size interpreter (BIO- openem regimen is changed to intermittent infusion dosing MIC V3 System; Giles Scientific, Santa Barbara, Calif).8 and is adjusted based on renal function upon transfer from the The main outcomes assessed in this study were assigned ICU.4 The intermittent group (February 1 to July 31, 2008) in- before study implementation and included duration of ventilator cluded patients who received conventional intermittent dos- support in intubated patients, ICU length of stay, hospital length ing (piperacillin-tazobactam 3.375 g intravenously [IV] every of stay, and in-hospital mortality. A pharmacoeconomic assess- 6 hours or meropenem 500 mg IV every 6 hours) with the ment of antibiotic costs was also performed and was based on the 30-minute bolus infusions. The prolonged group (August 1, July 2009 average wholesale price of the antibiotic used. 2008, to January 31, 2009) included patients who received prolonged infusion (piperacillin-tazobactam 3.375 g IV every Statistical Analysis 8 hours infused over 4 hours or meropenem 500 mg IV every Multiple linear regression analysis was used to compare 6 hours infused over 3 hours). The dosing intervals are ad- mean hospital length of stay, ICU length of stay, and ventilator justed based on renal function (Table 1). The piperacillin- days across study groups, adjusting for potential confounding tazobactam and meropenem dosing in patients with normal renal associated with differences in underlying patient risk char- function was adopted from Lodise et al.4 To verify the mer- acteristics between study groups. The estimates for these com- openem renal dosing, a Monte Carlo simulation revealed that parisons were adjusted for age, severity of illness (APACHE II), 500 mg IVevery 6 hours infused over 3 hours attained 40% f T 9 site of infection, other antibiotics while on study therapy, and MIC for 99% of isolates with MIC of 2 or less and 92% organisms recovered during treatment that were not amenable of isolates with an MIC of 4 for patients with an estimated to primary therapy. Similarly, multiple logistic regression was creatinine clearance of 40 mL/min (personal communication, used to analyze in-hospital death, adjusting for these potential G. Drusano, MD, March 24, 2008). The choice of antibiotics confounders. Total hospital length of stay, ICU length of stay, was decided by the prescriber based on patient history and risk and ventilator days may be nonnormal and right-skewed, thereby factors for resistant pathogens. If a patient was admitted to this violating the underlying distributional assumptions of classic ICU more than once during a single hospital admission, only the statistical approaches (eg, ordinary least squares). To account patient’s initial ICU stay was included in the study. Also, any for possible violations of distributional assumptions, therefore, patient whose Acute Physiology and Chronic Health Evaluation we estimated standard errors and confidence intervals (CIs) using a nonparametric bootstrapping approach with bias cor- II (APACHE II) score could not be determined based on the 9 medical record data available was excluded from the study rection and acceleration. Nonparametric bootstrapping does not analysis. rely on assumptions about the sampling distribution of the es- Data were collected from patients’ medical records using a timate and has favorable small sample properties resulting in more accurate type I error rates and greater power for identify- structured data collection tool to document age, sex, primary 10,11 type of critical care patient (medical or surgical), APACHE II ing small effects. Descriptive statistical analyses were per- score, ICU length of stay, hospital length of stay, in-hospital formed using SPSS (version 16.0; SPSS Inc, Chicago, Ill), and multivariate analyses were performed using Stata (version 11; StataCorp, LP, College Station, Tex). Two-sided P G 0.05 was TABLE 1. Prolonged Infusion Dose Adjustments Based regarded as statistically significant. on Renal Function RESULTS Estimated CrCl Dose The 183 evaluated patients were identified through billing Piperacillin-tazobactam infused over 4 h records as having received study medications for at least Q20 mL/min 3.375 g IV every 8 h 72 hours in the ICU. Seventy-eight patients were in the inter- G20 mL/min 3.375 g IV every 12 h mittent group and 105 patients in the prolonged group. Evalu- Hemodialysis/peritoneal dialysis 3.375 g IV every 12 h ated patients were excluded for the following reasons: 31 were receiving continuous renal replacement therapy; 2 received both Meropenem infused over 3 h regimens; none were cystic fibrosis patients; 6 were readmitted Q 36 mL/min 500 mg IV every 6 h to the ICU; in 4, we were unable to obtain their medical records; Y 26 35 mL/min 500 mg IV every 8 h and 19 were still in the hospital at the time of study conclusion. 10Y25 mL/min 500 mg IV every 12 h A total of 121 patients were included in the study, with 54 in G10 mL/min 500 mg IV every 24 h the intermittent group and 67 in the prolonged group. In the Hemodialysis/peritoneal dialysis 500 mg IV every 24 h intermittent group, 18 patients (33%) received meropenem, CrCl indicates creatinine clearance. whereas 36 patients (67%) received piperacillin-tazobactam. Thirteen patients (19%) received meropenem in the prolonged

2 www.infectdis.com * 2011 Lippincott Williams & Wilkins

Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Infectious Diseases in Clinical Practice & Volume 00, Number 00, Month 2011 Prolonged vs Intermittent A-Lactam Infusions group, and 54 patients (81%) received piperacillin-tazobactam. Demographic characteristics were similar between both study TABLE 3. Outcomes groups (Table 2). The average patient APACHE II scores were T T Difference $ Means 25.6 6.8 and 24.4 6.5 for the intermittent infusion and the Intermittent Prolonged (95% CI) prolonged infusion patients (P = 0.237), respectively. After adjusting for confounding variables by linear regression, the Total days of 8.97 9.14 0.17 (j1.85 to 2.39) intermittent group received therapy for an average total of therapy,* 8.97 days compared with 9.14 days in the prolonged group mean days ($ 0.17, 95% CI, j1.85 to 2.39). In contrast, the patients in the Days of therapy 7.16 6.31 j0.85 (j2.14 to 0.28) prolonged group had a reduced number of days of antibiotic in ICU,* therapy in the ICU (6.31 days) when compared with the patients mean days in the intermittent group (7.16 days; $ j0.85; 95% CI, j2.14 Ventilator days,* 16.8 9.6 j7.2 (j12.4 to j2.4)† to 0.28) (Table 3). mean days In the multiple linear regression analysis of ventilated ICU length 15.3 10.7 j4.5 (j8.3 to j1.4)† patients, time spent on the ventilator was reduced by 7.2 days of stay,* mean days (95% CI, 2.4Y12.4) in the patients receiving prolonged infusion j j j compared with those in the intermittent group (Table 3). A re- Hospital length 30.9 22.4 8.5 ( 18.7 to 1.2)† Y of stay,* duction in total ICU length of stay of 4.5 days (95% CI, 1.4 8.3) mean days as well as a reduction in the total hospital length of stay of Mortality,* % 20.7% 12.4% Odds ratio, 0.54 (0.18Y1.66) 8.5 days (95% CI, 1.2Y18.7) was associated with the prolonged infusion group. The risk of in-hospital mortality compared be- Adjusted for age, severity of illness (APACHE II), site of infection, tween the two groups by multiple logistic regression resulted in other antimicrobials while on study therapy, and organisms recovered a 12.4% mortality risk with prolonged infusion versus 20.7% during treatment not amenable to primary therapy. with intermittent infusion (odds ratio, 0.54; 95% CI, 0.18Y1.66). *Includes 121 evaluable patients. Although this was not significant, the odds of dying in patients †Significant difference. from the prolonged group were 0.54 times lower than those in the intermittent group (Table 3). A similar percentage of culture proven (prolonged, 45%, There were a total of 162 documented infections at multiple sites vs intermittent, 44%) and clinically documented (prolonged, that were comparable between groups (Table 2). Table 2 depicts 55%, vs intermittent, 56%) infections were found in both groups. the gram-negative organisms isolated from culture. Sixty-nine isolates had MICs performed to piperacillin-tazobactam and/or meropenem, with 65 isolates (94%) found to be susceptible to either agent (median MIC = 0.5; range, G0.25Y14). Sixty-four TABLE 2. Demographic Characteristics percent (44/69) of our isolates had an MIC of 1 or less to both Intermittent Prolonged agents. Four isolates (6%) were found to be nonsusceptible to piperacillin-tazobactam, and none of the isolates were found to n = 54 n = 67 P be nonsusceptible to meropenem, according to the established 8 Age, average (SD) 60.4 (16.4) (18Y89) 58.4 (17.2) (18Y88) 0.527 MIC breakpoints for each antibiotic during the study period. (range), y Three of these isolates ( aeruginosa, Enterobacter Males, n (%) 33 (61) 46 (69) 0.444 cloacae, and ) were from the intermittent group, Patient type, % whereas 1 isolate (Enterobacter aerogenes) was from the pro- Medical 61 69 0.565 longed group. Patients with resistant isolates had their therapy Surgical 39 31 changed to appropriate antibiotics to cover the infecting organ- ism. (79%), ciprofloxacin (66%), and metronida- APACHE II score, 25.6 (6.8) (11Y39) 24.4 (6.5) (8Y41) 0.237 mean (SD) (range) zole (11%) were the most common concomitant antibiotics used with either piperacillin-tazobactam or meropenem, and Intubated patients 51 60 there were no statistical differences in use between treatment Presumed site of groups. Twelve patients (10%) did not receive any concomitant infection, % antibiotics. Pulmonary 40 53 There was a decrease in the DDD per 1000 patient days of Intra-abdominal 25 17 piperacillin-tazobactam in the ICU from 128.3 in the 6 months Blood 10 11 of intermittent infusion therapy to 115.4 during the first 6 months Pancreas 9 4 of the prolonged infusion in the ICU. There was also a decrease Urine 7 6 in the meropenem DDD per 1000 patient days from 58.6 during Gram-negative the intermittent infusion period to 34.5 in the prolonged infusion isolates, n period. Pseudomonas 15 13 aeruginosa DISCUSSION E. coli 23 5 The prolonged infusions of piperacillin-tazobactam and Klebsiella spp 9 0 meropenem were adopted into clinical practice in the medical Citrobacter spp 3 5 and surgical ICU after promising results in Monte Carlo simu- Enterobacter spp 3 4 lations were followed by improved outcomes in critically ill 2,4 Haemophilus spp 0 2 patients. In this retrospective analysis of a newly imple- mented prolonged infusion strategy, we demonstrated signifi- Other 2 2 cantly improved outcomes in reduced ventilator days and ICU

* 2011 Lippincott Williams & Wilkins www.infectdis.com 3

Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Dow et al Infectious Diseases in Clinical Practice & Volume 00, Number 00, Month 2011 and hospital length of stay, as well as a trend toward lower macokinetics in critically ill patients, which may give the pro- mortality with prolonged infusion. longed infusion strategy of A-lactams an advantage over the In Monte Carlo simulation, the probability of attaining intermittent infusion in this population. 50% f T 9 MIC for the prolonged infusion of piperacillin- Relatively few studies have evaluated the clinical outcomes tazobactam was 92% at an MIC of 16 mg/L and 100% at MICs of the prolonged infusion of piperacillin-tazobactam and mer- of less than 16 mg/L.4 When piperacillin-tazobactam was openem. This study is a preprotocol and postprotocol study that dosed as a 30-minute bolus infusion every 6 hours, the proba- validates the previously reported positive outcomes in this bility of attaining 50% f T 9 MIC was greater than 90% only population.2 Although we did not demonstrate a significant for an MIC of 1 mg/L. Lodise and colleagues2 reported that difference in mortality in our patient population, several differ- critically ill patients with only P. aeruginosa infections who ences in our study should be discussed. First, we did not limit had APACHE II scores of 17 or greater receiving the pro- our study only to patients with culture-proven P. aeruginosa longed infusion of piperacillin-tazobactam had a significantly infections. Only 11 patients (6 in intermittent group and 5 in lower 14-day mortality rate when compared with those who re- the prolonged group) had documented P. aeruginosa infec- ceived the bolus intermittent infusion (12.2% vs 31.6%, re- tions. Only 1 of these 11 patients, from the intermittent group, spectively; P = 0.04), as well as a significantly lower median had a P. aeruginosa isolate that was reported to be resistant to duration of hospital stay (21 vs 38 days, respectively; P = 0.02). piperacillin-tazobactam. In this medical and surgical ICU for Differences in duration of ventilation and ICU length of stay 2008, 13% of all the gram-negative isolates (not just from our outcomes between the 2 groups were not reported.2 study) were nonsusceptible to meropenem (all P. aeruginosa), Meropenem was also included in the prolonged infusion whereas 21% were nonsusceptible to piperacillin-tazobactam protocol based on promising but limited clinical data. A 500-mg (Acinetobacter baumannii, E. cloacae, E. coli, Klebsiella oxytoca, dose of meropenem administered by prolonged infusion over 3 , and P. aeruginosa). We included hours has been shown to result in mean drug exposures (%f T 9 patients with both documented and suspected infections, con- MIC) above the MICs of 4 and 1 mg/L of 47.27% and 71.44% of sistent with clinical practice in patients admitted to the ICU. the dosage interval, respectively. This was significantly greater The patients with suspected infections had their antibiotics than that obtained with a 1000-mg bolus dose at the same MICs, continued by their attending physicians secondary to clinical 42.5% and 67.04%, respectively (P G 0.05).6 signs and symptoms. A study by Nicasio and colleagues7 compared The findings of our study differ from those reported by pharmacodynamic-based antibiotic dosing in patients treated Patel and colleagues,15 which did not observe a significant dif- for ventilator-associated pneumonia (VAP) with historical con- ference in 30-day mortality or median length of hospital stay trols. In 21.3% of these patients, meropenem 2 g IVevery 8 hours by comparing patients receiving intermittent infusion to pro- infused over 3 hours was given. The pharmacodynamic-based longed infusion piperacillin-tazobactam. Contrary to this previ- dosing regimens significantly decreased infection-related mor- ous study, our investigation focused solely on critical care tality by 13.1%, infection-related length of stay by 14.4 days, patients, who may benefit from improved coverage with pro- and superinfections by 19.1%. There was no difference in longed infusions due to the discussed changes in antimicrobial 28-day mortality, ICU days after VAP, ventilator duration after and pharmacodynamics that occur in this VAP, or total hospital length of stay for the historical versus complex population. This may account for the difference in pharmacodynamic-based dosing patients. The primary prolonged outcomes. Overall, disease severity was low in their study infusion A-lactam in this study was in approximately population. The mean APACHE II scores were only 10.5 and 75% of patients, which was not evaluated in our study. 10.9 for the intermittent and prolonged patients, respectively. The results of our study support the use of the prolonged Only 4% (intermittent group) and 7% (prolonged group) of infusions of piperacillin-tazobactam and meropenem in critically their patients were in the ICU at the time of culture collection, ill patients. Time spent on the ventilator, total ICU length of and their overall mortality was much lower (intermittent group, stay, and hospital length of stay were also significantly de- 5%; prolonged group, 4%), as was their median hospital length creased in this study. A potential theory as to why the pro- of stay (8 days for both groups). longed infusion strategy may be beneficial in critically ill The prolonged infusion of piperacillin-tazobactam is as- patients may be due to changes in pharmacokinetics in this sociated with the benefit of cost savings by reducing the dosing population. These physiological changes that can occur in these frequency from every 6 hours to every 8 hours. Based on av- patients may lead to inadequate concentrations of A-lactam erage wholesale price, the 54 patients who received the pro- antibiotics and decreased efficacy.12,13 In sepsis, patients often longed infusion of piperacillin-tazobactam during the 6-month have an increased volume of distribution due to leaky capillar- period of the study, and assuming a 7- to 10-day treatment ies and altered protein binding, as well as an increased clearance period, $8765 to $12,522 were saved by using the prolonged due to an increased cardiac index.12 The increased clearance of infusion dosing in this ICU. The reduction in the DDD for A-lactam antibiotics has been demonstrated in this population.12 piperacillin-tazobactam was despite an increase in the actual Hypoalbuminemia is also common in critically ill patients, days of use by 76 days, which reflects significant cost sav- resulting in lower than anticipated concentrations of drug in the ings with the use of the prolonged infusion of piperacillin- blood. This can create an increase in the free drug concentration, tazobactam. In addition, our study also demonstrates cost which in turn may reduce the half-life of the drug.13 Although a savings due to the reduction in ventilator days and both ICU and statistically significant difference in in-hospital mortality was total hospital length of stay. Based on previously published not shown in this study, the lower risk of mortality in patients literature on the estimated cost of an ICU day, the approximate who received the prolonged infusion may represent a promising 4-day decrease in ICU length of stay in our study resulted in trend. We did not find a significant difference in total days of cost savings of $14,000 per patient in ICU stay alone.16 The therapy or days of therapy in the ICU as a result of reliance on reduction in DDD for meropenem can be accounted for by a evidence-based guidelines for duration of therapy, such as when decrease in the total days of use in the ICU during this period treating VAP.14 We believe the large difference in ventilator (115 days), as the meropenem dosing regimen did not change, days to be secondary to the above explained change in phar- just the infusion time.

4 www.infectdis.com * 2011 Lippincott Williams & Wilkins

Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Infectious Diseases in Clinical Practice & Volume 00, Number 00, Month 2011 Prolonged vs Intermittent A-Lactam Infusions

Several limitations of the study should be discussed. The 3. Craig WA. Pharmacokinetic/pharmacodynamic parameters: rationale study was retrospective in nature, and data are limited to the for antibacterial dosing of mice and men. Clin Infect Dis. 1998;26:1Y10. depth and the accuracy of the documented medical record. 4. Lodise TP, Lomaestro BP, Drusano GL. Application of antimicrobial The study population was confined to a single-center medical/ pharmacodynamic concepts into clinical practice: focus on Q-lactam surgical ICU. A relatively small number of patients received antibiotics. Insights from the Society of Infectious Diseases Pharmacists. meropenem, which consequently makes it difficult to draw con- Pharmacotherapy. 2006;26(9):1320Y1332. clusions on the efficacy of prolonged infusion of meropenem 5. Kim A, Sutherland CA, Kuti JL, et al. Optimal dosing of individually. Patients were included with documented as well as piperacillin-tazobactam for the treatment of suspected infections, and it was not limited to patients with infections: prolonged or continuous infusion? Pharmacotherapy. only documented, culture-proven gram-negative infections. This 2007;27:1490Y1497. may give the study improved external validity as piperacillin- 6. Jaruratanasirikul S, Sriwiriyajan S. Comparison of the tazobactam or meropenem is often used as empiric broad- pharmacodynamics of meropenem in healthy volunteers following spectrum coverage in critically ill patients and then continued administration by intermittent infusion or bolus injection. despite the absence of a culture-proven infection as the patient J Antimicrob Chemother. 2003;52:518Y521. may still show clinical signs of infection. There was also a relative 7. Nicasio AM, Eagye KJ, Nicolau DP, et al. Pharmacodynamic-based lack of gram-negative bacteria with high MICs during our study clinical pathway for empiric antibiotic choice in patients with period. However, this is consistent with our local ICU antimi- ventilator-associated pneumonia. J Crit Care. 2010;25:69Y77. crobial susceptibility patterns over several years. Antibiotic regi- mens used before the initiation of piperacillin-tazobactam and 8. Clinical and Laboratory Standards Institute. Performance Standards for meropenem were not collected. This could have influenced out- Antimicrobial Susceptibility Testing. Nineteenth Informational comes if patients were not treated appropriately, although only Supplement. CLSI Document M100-S19. Wayne, PA: Clinical and Laboratory Standards Institute; 2009. 9% of the intermittent group and 7% of the prolonged group had positive cultures drawn 48 hours or more before the initiation 9. Efron B. Better bootstrap confidence intervals. J Am Stat Assoc. of study antibiotics. 1987;82(397):171Y185. The length of the study period was also limited and there- 10. Yung Y, Chan W. Statistical analyses using bootstrapping: concepts fore may have had insufficient power to detect a difference in and implementation. In: Hoyle R, ed. Statistical Strategies for Small overall mortality. We do acknowledge that other threats to va- Sample Research. Thousand Oaks, CA: Sage Publications; 1999. lidity cannot be ruled out by the study design, such as regression 11. Barber JA, Thompson SG. Analysis of cost data in randomized trials: to mean and the Hawthorne effect in the prolonged period. an application of the non-parametric bootstrap. Stat Med. In conclusion, this study shows that the use of the pro- 2000;19(23):3219Y3236. longed infusions of piperacillin-tazobactam or meropenem 12. Roberts JA, Lipman J. Antibacterial dosing in intensive care: could potentially improve clinical and pharmacoeconomic out- pharmacokinetics, degree of disease and pharmacodynamics of sepsis. comes in critically ill populations. More rigorous scientific Clin Pharmacokinet. 2006;45:755Y773. studies in critically ill patients in a prospective manner, espe- 13. Diaz E, Ulldemolins M, Lisboa T, et al. Management of cially further studies with meropenem, at other institutions ventilator-associated pneumonia. Infect Dis Clin N Am. 2009;23: may support our findings and promote further implementation 521Y533. of this strategy. 14. American Thoracic Society. Infectious Diseases Society of America. Guidelines for the management of adults with hospital-acquired, REFERENCES ventilator-associated, and healthcare-associated pneumonia. Y 1. Mattoes HM, Kuti JL, Drusano GL, et al. Optimizing antimicrobial Am J Respir Crit Care Med. 2005;171:388 416. pharmacodynamics: dosage strategies for meropenem. Clin Ther. 15. Patel GW, Patel N, Lat A, et al. Outcomes of extended infusion 2004;26:1187Y1198. piperacillin/tazobactam for documented gram-negative infections. Y 2. Lodise TP, Lomaestro B, Drusano GL. Piperacillin-tazobactam for Diagn Microbiol Infect Dis. 2009;64:236 240. Pseudomonas aeruginosa infection: clinical implications of an 16. Dasta JF, McLaughlin TP, Mody SH, et al. Daily cost of an extended-infusion dosing strategy. Clin Infect Dis. 2007;44: intensive care unit day: the contribution of mechanical ventilation. 357Y363. Crit Care Med. 2005;33:1266Y1271.

* 2011 Lippincott Williams & Wilkins www.infectdis.com 5

Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.