Economic assessment of doripenem versus imipenem in the treatment of ventilator-associated pneumonia

Abstract

Background: Ventilator-associated pneumonia (VAP), the most common nosocomial infection in critically ill patients, is associated with significantly longer duration of mechanical ventilation, and increased mortality, hospital days, and health-care costs. A previously published prospective, randomized study established the noninferiority of intravenous (IV) doripenem versus IV imipenem/cilastatin (‘imipenem‘) for VAP. This study compares the economic outcomes of IV therapy with doripenem versus imipenem as first-line treatment for VAP.

Methods: A decision-analytic model of inpatient care and outcomes for VAP was used to estimate costs associated with VAP treatment. The model calculates total hospital costs, comprising costs of initial and concomitant therapy, and costs associated with mechanical ventilation, intensive care unit stays, and total days in hospital.

Results: Total treatment costs for doripenem were $10,630 lower than for imipenem ($71,259 vs. 81,889), driven primarily by differences in costs of mechanical ventilation ($45,224 for doripenem, $57,348 for imipenem). Probabilistic sensitivity analyses found doripenem consistently cost saving versus imipenem in 1,000 simulations. Study limitations include use of a simple model to represent a complex disease process and reliance on trial data that may not reflect real-world care and outcomes.

Conclusions: Doripenem is a cost saving first-line treatment for VAP versus imipenem while providing an equivalent rate of cure.

Key words::

• cost effectiveness
• doripenem
• economic model
• imipenem
• ventilator-associated pneumonia

Introduction

Ventilator-associated pneumonia (VAP), the most common nosocomial infection in critically ill patients, is associated with high mortality rates, increased hospital stays, significantly longer duration of mechanical ventilation, and increased healthcare costs^1–4. VAP is the most frequent infection acquired in the intensive care unit (ICU), accounting for 45% of all infections in European ICUs⁵. Between 10 and 20% of US patients receiving more than 48 hours of mechanical ventilation will develop VAP, and mortality is twice as high among critically ill patients who develop VAP compared with those who do not¹; the VAP-attributable mortality among VAP patients who die in hospital in the US is estimated to be between 30 and 55%⁶. VAP is estimated to increase hospital stays by 4–13 days leading to additional costs of $20,000–40,000 per case⁵; the growing presence of antimicrobial resistance in the ICU has contributed to substantial increases in overall healthcare costs^6–8.

A carbapenem with antipseudomonal activity is generally recommended for initial treatment of VAP, especially in patients at risk for multidrug-resistant pathogens⁹. Imipenem, a carbapenem commonly used to treat hospital-based infections including VAP, is effective against Gram-negative pathogens. Doripenem, an injectable, broad-spectrum carbapenem (β-lactam) that has activity against key Gram-negative and Gram-positive pathogens responsible for VAP, is approved in the US for complicated intra-abdominal and complicated urinary tract infections and is under FDA review for the treatment of nosocomial pneumonia (NP)/VAP^10,11; it is approved for all three indications in many European countries. The antibacterial spectrum of doripenem is similar to that of imipenem and meropenem, and doripenem exhibits lower in vitro minimal inhibitory concentration for 90% of isolates (MIC₉₀) against Pseudomonas aeruginosa, one of the leading pathogens in VAP, than these comparators¹².

A previously published multicenter, prospective, randomized, open-label, phase 3 study¹³ established the noninferiority of intravenous (IV) doripenem versus IV imipenem/cilastatin (referred to as imipenem throughout the document) in the treatment of VAP. Patients (n=525) were randomized 1:1 to receive doripenem or imipenem; the primary outcome was the clinical cure rate at the test-of-cure (TOC) visit. Duration of study drug and duration of concomitant medication were collected in the trial, as were duration of hospital stay, days in the ICU, and days on mechanical ventilation. Trial participants on both doripenem and imipenem experienced cure rates of 63.9% and mortality rates were similar in the two groups. However, patients on doripenem had significantly shorter hospitalizations and significantly fewer days on mechanical ventilation compared with imipenem patients¹³.

In evaluating comparable treatments, it has become increasingly important to understand not only efficacy but the costs of interventions aimed at reducing mortality and morbidity of critically ill patients. Countries such as the UK, Canada, and Australia require formal submission of economic data to gain formulary reimbursement¹⁴. The use of economic analyses to aid treatment and reimbursement decisions has become widely accepted in the US as well¹⁵. Therefore a model was developed to estimate the economic costs associated with doripenem versus imipenem in the treatment of VAP in the US. A primary source of data for estimation of this model was the aforementioned VAP trial¹³.

Methods

Techniques of decision analysis were used to construct and estimate a model of the treatment and outcomes for patients treated for VAP over the course of hospitalization. The clinical decision of interest was the choice of empiric antimicrobial treatment for VAP. All patients were assumed to enter the model with a diagnosis of VAP while receiving treatment in the ICU on mechanical ventilation. The target population for the model reflects the clinically modified intent-to-treat study population in the VAP trial¹³. Participants in the VAP trial were adult patients (mean age, 52 years) with newly diagnosed pneumonia who had received mechanical ventilation for more than 24 hours, or who had been weaned from mechanical ventilation within 72 hours.

Total inpatient costs, including costs associated with initial treatment, concomitant therapy, mechanical ventilation, ICU stays, and total LOS in hospital, were calculated in the model. Model analyses were conducted from the payers' perspective and, therefore, include only direct medical costs. Costs for patients with VAP were tracked from treatment initiation until hospital discharge or death. Because the two comparators were found to be equally safe and efficacious in the trial, it was decided to not assess differences in treatment outcomes or perform a cost-effectiveness analysis – this analysis therefore is in the realm of cost-minimization analysis.

Model calculations

Total inpatient costs for treatment of VAP were calculated by summing the costs of initial therapy, concomitant therapy, and hospitalization. Costs of initial therapy were calculated by multiplying the mean days on study therapy for patients in each treatment group by the average doses per day and cost per dose of the therapy received. For analyses of concomitant therapy, amikacin and vancomycin were considered, which were recommended in the trial for patients suspected as having Pseudomonas aeruginosa or methicillin-resistant Staphylococcus aureus, respectively. Costs of concomitant therapy with vancomycin and amikacin were calculated by multiplying the estimated proportion of patients receiving each therapy by the average duration of therapy by the estimated average doses per day and cost per dose of the therapy received, and then summing across both concomitant therapies.

Costs of hospitalization were calculated by estimating the mean number of hospital days spent in one of three inpatient settings: (1) the ward, (2) ICU, and (3) ICU on mechanical ventilation, and assigning separate costs to time spent in each setting. The estimated daily cost for each setting was multiplied by the mean number of days in the setting, and summed across all three inpatient settings to estimate total inpatient costs.

Model estimation

The model was estimated by assigning values to all model inputs for resource utilization and unit costs. Clinical trial data provided estimates of the proportion of patients receiving concomitant medications; mean days on initial therapy, mean days on concomitant medications, and mean length of stay (LOS), including time spent in ICU with and without mechanical ventilation (Table 1).

Table 1. Utilization parameters and data sources.

Model parameter	Doripenem estimate (SE*)	Imipenem estimate (SE*)	Data SOURCE
Mean days on study drug	7.73 (3.07)	8.09 (3.06)	Merchant et al (2008)^13
Proportion of patients receiving:
Amikacin	0.112 (0.020)	0.159 (0.023)	Merchant et al (2008)^13
Vancomycin	0.281 (0.028)	0.294 (0.029)	Merchant et al (2008)^13
Mean days on:
Amikacin	4.29 (0.71)	4.35 (0.58)	Data on file at Johnson & Johnson
Vancomycin	5.69 (1.10)	4.89 (0.65)	Data on file at Johnson & Johnson
Length of stay (mean days):
Total	24.07 (0.81)	27.46 (0.85)	Merchant et al (2008)^13
ICU	14.63 (0.64)	16.19 (0.71)	Merchant et al (2008)^13
Mean days on mechanical ventilation	10.22 (0.62)	12.96 (0.81)	Merchant et al (2008)^13

ICU: intensive care unit.

*SE: Standard errors were used in the probabilistic sensitivity analysis.

Cost inputs and data sources are summarized in Table 2. Drug acquisition costs (wholesale) for the initial therapy and concomitant medications were extracted from Analy$ource, based on the National Drug Data File (NDDF; First DataBank, Inc.)¹⁶. Based on the dosing regimens specified in the VAP clinical trial protocol, patients receiving doripenem were assumed to receive 500 mg every 8 hours and patients receiving imipenem either 500 mg every 6 hours or 1,000 mg every 8 hours. The doripenem cost per dose was estimated at $38.33; it was assumed patients would receive three doses per day, resulting in a cost per day of $114.99. The imipenem cost per dose was estimated at $32.65. Approximately 62% of patients received 500 mg every 6 hours, resulting in a cost per day of $130.60, and 38% of patients received 1,000 mg every 8 hours, resulting in a cost per day of $195.90. The weighted average of the two imipenem dosages was calculated to obtain a cost per day of $155.12. The average daily dose for amikacin was obtained from the VAP trial protocol, whereas the average daily dose for vancomycin was obtained from the prescribing information.

Table 2. Cost parameters and data sources.

Cost per day	Estimate (SE*)	Data source
Initial therapy:
Doripenem	$114.99 (11.5)	Extracted from Analy$ource, based on the National Drug Data File (NDDF) and copyrighted by First DataBank, Inc. (2007)^16
Imipenem	$155.12 (15.5)
Concomitant therapy:		Extracted from Analy$ource, based on the NDDF and copyrighted by First DataBank, Inc. (2007)^16
Amikacin	$10.63 (1.1)
Vancomycin	$13.91 (1.4)
Hospital setting:
Ward	$1,116 (111.6)	Paladino et al (2007)^17
ICU without mechanical ventilation	$3,307 (330.7)	Dasta et al (2005)^18
ICU with mechanical ventilation	$4,425 (442.5)	Dasta et al (2005)^18

ICU: intensive care unit.

*SE: Standard errors were used in the probabilistic sensitivity analysis.

Inpatient costs by hospital setting were estimated from published literature. The cost per day in ward was obtained from Paladino et al¹⁷; the ICU costs per day with and without mechanical ventilation were obtained from Dasta et al¹⁸, and reflect the mean cost per day for patients admitted to the ICU for medical reasons other than surgery or trauma. Since patients enrolled in the VAP trial were admitted to the hospital prior to acquiring VAP, the cost estimate from Dasta et al¹⁸ was used for days spent in ICU or in ICU on mechanical ventilation for patients transferred to the ICU, rather than that for patients admitted directly to the ICU. All analyses were either calculated in 2006 US dollars or converted to 2006 dollars using the Consumer Price Index (CPI) for medical care services.

Economic analyses

The cost savings of doripenem versus imipenem in the treatment of VAP were calculated in terms of the incremental treatment cost per patient. One-way sensitivity analyses were conducted to evaluate the robustness of the model results to plausible changes in input parameters. To quantify the uncertainty in the estimates of the cost of treating VAP due to uncertainty in the estimation of input parameters, a probabilistic sensitivity analyses using second-order Monte Carlo simulations was performed¹⁹. To set up the Monte Carlo analyses, uncertainty in key model parameters was characterized by probability distributions around each of their midpoint estimates assuming gamma distributions for costs, normal distributions for lengths of stay and duration of treatment and beta distribution for proportions. Standard errors were derived from the clinical trial data for all utilization parameters (Table 1). No distribution was available for drug costs, and the reported distributions around the inpatient costs derived from large database analyses were thought to understate the uncertainty of the estimates; therefore, a distribution of 90–110% of all base-case cost values was assigned (Table 2). A random number generator was used to ‘draw’ parameter values from each distribution, and these values were run through the model to generate cost estimates. This process of drawing parameters and running the model was repeated 1,000 times.

Results

Economic analyses

Total treatment costs for doripenem were $10,630 lower than for imipenem patients (Figure 1). The cost of hospital days on mechanical ventilation comprised the bulk of the total treatment costs, and the difference in overall treatment costs for VAP patients was driven primarily by differences in costs associated with days on mechanical ventilation, which were $12,124 lower for doripenem patients ($45,224) than for imipenem patients ($57,348).

Figure 1. Per-patient cost of hospitalization among VAP patients, by initial treatment.

Sensitivity analyses

The cost estimates generated by the model are most sensitive to estimates of total inpatient days and days in ICU on each therapy; however, doripenem remains cost saving versus imipenem even when total hospital days and days in ICU are set equal for both therapies. Among the unit cost inputs, the model is most sensitive to the assumed daily cost of mechanical ventilation; however, even if there is no additional cost for ICU days on mechanical ventilation, doripenem remains cost saving. Results are relatively insensitive to drug cost and utilization assumptions of both initial and concomitant therapy: even assuming doripenem costs in excess of $500/day (> 4 times the base case), doripenem is cost saving versus imipenem.

Results of the probabilistic sensitivity analysis show that doripenem was consistently cost saving versus imipenem in 1,000 simulations (95% credible interval cost savings of $5,100–16,500).

Discussion

VAP is associated with high morbidity and mortality, and increasing prevalence of multidrug-resistant bacterial strains highlights the need for new therapies to combat infections caused by serious pathogens such as P. aeruginosa. In a recent phase 3 trial of empiric treatment for VAP, the clinical cure rate for doripenem, a new investigational carbapenem, was found to be clinically noninferior to imipenem. Data from this trial, showed that total hospital LOS, ICU LOS, and time on mechanical ventilation were consistently lower for patients treated with doripenem in comparison with patients treated with imipenem¹³. The incidence of treatment-emergent adverse events related to study drug and the mortality rates also were balanced between the imipenem and doripenem treatment arms, and there were fewer total ICU readmissions for any cause among the doripenem subjects compared with the imipenem subjects.

The purpose of the current study was to examine the economics of doripenem versus imipenem in the treatment of VAP. Overall, there is a dearth of economic literature on the management of the critically ill patient¹³: very limited cost analyses of doripenem have been conducted and only a handful of studies have assessed the cost or cost effectiveness of other carbapenems in the treatment of nosocomial pneumonia^20–24. To address this information gap, modeling techniques were used to analyze the cost of first-line treatment of VAP with doripenem versus imipenem. The model used clinical trial data in addition to other publicly available sources to estimate costs for each strategy. The results show that inpatient costs of VAP treatment with doripenem are lower than imipenem ($71,259 vs. 81,889 per patient), and clinical trial results show the two treatments to be equally effective in terms of clinical cure. Doripenem therefore may be considered the cost-effective strategy choice because selecting doripenem versus imipenem results in lower VAP treatment costs without any corresponding trade-off in effectiveness.

The estimates of the high cost of VAP are consistent with previous studies. For example Shorr et al⁴ conducted a retrospective analysis of inpatient records and reported that, under the conservative assumption that the daily cost of ICU care in the US is $2,000, the presence of VAP increases costs by at least $10,000–50,000 per inpatient stay. Rello et al², in a retrospective matched cohort study using data from a large US inpatient database, found that development of VAP was associated with an increase of more than $40,000 in average hospital charges per patient.

The results of this study are subject to a number of limitations. First, the decision-analytic model is necessarily a simplified representation of the disease and treatment process and cannot include all possible strategies and outcomes. The authors recognize that acutely ill patients on mechanical ventilation are a heterogeneous group, and estimating costs for VAP patients as a whole may mask important underlying differences in risks and outcomes between patients with diverse admitting diagnoses. In addition, the authors note that the data available from the VAP trial were collected primarily for clinical research purposes, and therefore were not ideally suited to performing an economic analysis. In particular, clinical cure was only systematically evaluated at the test-of-cure (TOC) visit which occurred after most patients were discharged from the ICU and removed from mechanical ventilation; therefore, although the differences in days on mechanical ventilation were assumed to be due to differences in timing of cure, this relationship could not be confirmed from the available data. In addition, due to limitations in the available data, the model does not take into account potential real-world patterns of therapy switching and de-escalation that may affect both treatment costs and outcomes. The authors also note that the VAP trial was not blinded, due to the fact that there were different dosing regimens for imipenem (three times or four times a day) and doripenem was used as a 4-hour infusion versus a 1-hour infusion for imipenem; therefore, it is possible that clinician awareness of treatment assignment may have influenced treatment decisions, including the decision to terminate mechanical ventilation or to discharge the patients from the ICU or the hospital, however, several restrictions were in place to minimize bias. Although study drugs were administered in an unblinded manner, patient evaluability for primary and secondary analyses was blinded. The statistical team was blinded to treatment until database lock, and the medical team responsible for conduct of the study and determining patient evaluability was also blinded during the review process.

Because the model was estimated primarily using data from a clinical trial with strict inclusion and exclusion criteria and protocol-driven treatment¹³, care should be used in generalizing the results of this study to other settings and populations. Costs used in this analysis were taken from published data and standard sources; the extent to which they reflect the true costs of administering medical care is unknown. In particular, the use of wholesale acquisition cost (WAC) to estimate drug acquisition costs likely overstates the actual cost of drugs to hospital pharmacies. Furthermore, this study was conducted from a third-party payer perspective rather than a societal perspective, and as such does not include such costs of VAP as lost productivity, caregiver time, transportation, and other unreimbursed expenses. Inclusion of indirect costs presumably would have added substantially to the total cost burden. The model also is limited to the duration of initial hospitalization and does not take into account longer-term consequences of treatment failure, post-discharge costs, or costs of rehospitalization, which may be substantial. And finally, utility values to calculate quality-adjusted life-years (QALYs) were not collected in the VAP trial. Given the acute nature of VAP and the very small and statistically non-significant difference in mortality rates between the study groups, assessing effectiveness in terms of QALYs was considered inappropriate and was not undertaken¹⁵.

Conclusion

While clinical endpoints from randomized trials continue to be of primary importance when evaluating comparable treatments, economic models can be highly informative to payers and other healthcare decision makers when the treatment alternatives offer similar clinical benefits. Data from a recent phase 3 trial found doripenem to be clinically noninferior to imipenem in the treatment of VAP. The current study demonstrates that doripenem is cost saving versus imipenem in VAP when taking into consideration the total cost of inpatient care, including antibiotics, mechanical ventilation, and length of stay in the hospital. The economic findings from this paper, combined with previous clinical findings, suggest that doripenem may be considered an alternative for empirical treatment of VAP.

Transparency

Declaration of funding: This study was funded by Johnson & Johnson Pharmaceutical Services, LLC.

Declaration of financial/other relationships: L.J.M., V.P., K.D., D.T., and M.C.W. have disclosed that they are paid consultants of Johnson & Johnson. S.M., K.A. and M.I. have disclosed that they are employees of Johnson & Johnson. D.P. has disclosed that he has received honoraria for serving on Johnson & Johnson's Advisory Board in the past, but has received no honoraria for this study.

The JME peer reviewers 1 and 2 have not received an honorarium for their review work on this manuscript. Both have disclosed that they have no relevant financial relationships.

Acknowledgements: The authors would like to acknowledge the assistance of Victoria Porter, i3 Innovus, for her help with editorial and manuscript preparation; and Chris Gast, Johnson & Johnson, for his help with statistical analysis.

Notes