Introduction

Staphylococcus aureus is the most common cause of infection following solid organ transplantation. Lung transplant recipients, in particular, are highly susceptible to S. aureus infections due to the need for especially robust immunosuppression, lengthy hospital and intensive care unit (ICU) stays, frequent colonization of the upper respiratory tract, and impaired cough and mucociliary clearance due to denervation of the transplanted lung [1]. S. aureus infections were diagnosed within the first 90 days of lung or heart-lung transplantation in 18 % (109/596) of our patients at the University of Pittsburgh Medical Center (UPMC) between 2006 and 2010 [1]. Independent risk factors for S. aureus infections were positive S. aureus cultures from the airways of explanted lungs (p = 0.003) and positive methicillin-resistant S. aureus (MRSA) cultures from recipients’ nares (p < 0.0001). The data suggest that S. aureus screening and decolonization strategies, if effective at preventing infections and their sequelae, would have major impact on outcomes following lung transplantation.

Studies of S. aureus screening and decolonization have not been undertaken in lung transplant recipients. In general surgical populations, decolonization of S. aureus nasal carriers reduced the rate of post-operative S. aureus infections by 40–60 % [2, 3]. Patients undergoing cardiothoracic surgery are considered high-impact populations for screening and decolonization because the potential consequences of S. aureus surgical site infections (SSIs) in these patients are particularly deleterious [4]. Data on the value of active S. aureus surveillance and decolonization following cardiothoracic and other surgeries are conflicting. Studies indicate that such strategies can prevent MRSA SSIs, but the impact on more severe infections like mediastinitis, pneumonia or bacteremia is unclear [4, 5]. Moreover, screening and decolonization strategies are resource-intensive and expensive, and their economic value remains inconclusive [6].

Studies employing simulation models have shown that S. aureus screening and decolonization of hemodialysis and orthopedic, vascular, and cardiac surgery patients provide economic value to both hospitals and third-party payers [7–10]. These findings, however, are not necessarily applicable to lung or other transplant populations. In this study, we developed a stochastic decision analytic simulation model to determine the cost-effectiveness of routine peri-operative S. aureus screening and decolonization of lung and heart-lung transplant recipients from hospital and third party payer perspectives. Sensitivity analyses varied key parameters such as S. aureus colonization prevalence, probability of infection, and decolonization efficacy.

Methods

Clinical data and protocols

Clinical data were collected during a retrospective cohort study of S. aureus infections among consecutive lung and heart-lung transplant recipients at UPMC from January 2006 to December 2010 [1]. Our standard immunosuppression, antimicrobial and surveillance bronchoscopy regimens have been published. The study was reviewed and approved by the University of Pittsburgh Institutional Review Board (IRB).

A nares culture swab for MRSA was obtained on all patients admitted to the cardiothoracic ICU (CT-ICU) after transplantation. Prior to 2010, there were no recommendations for S. aureus decolonization at our center. Beginning in 2010, it was recommended that CT-ICU patients who were nares culture-positive for MRSA or had S. aureus recovered from the airways of explanted lungs at transplant (i.e, recipient sterility culture-positive) receive twice-daily intranasal 2 % mupirocin ointment and daily chlorhexidine-gluconate soap (40 mg/mL) baths for 5 days [2]. Adherence with this recommendation was not enforced during the study period.

Definitions

S. aureus infections within the first 90 days post-transplant were defined according to International Society of Heart and Lung Transplantation (ISHLT) consensus criteria [11], and included both methicillin-susceptible S. aureus (MSSA) and MRSA infections. Patients with positive recipient sterility cultures for S. aureus or positive nares cultures for MRSA were defined as colonized. S. aureus infection rates in the absence of decolonization and after decolonization were based on 2006–2009 and 2010 UPMC data, respectively. Decolonization efficacy was the reduction in infection rates after the mupirocin-chlorhexidine protocol was recommended.

Model structure and inputs

The model (Fig. 1) was developed in TreeAge Pro 2009 (Williamstown, MA, USA). Each transplant recipient entering the model was either screened or not screened post-operatively, within 7 days of surgery. Those with a positive screening test underwent decolonization (as described above). Patients then had a probability of developing an infection due to MSSA or MRSA; those who underwent decolonization had a reduced risk of infection, based on clinical data. Additional scenarios explored the effects of decolonization by attenuating the probability of S. aureus infection (using data assuming no decolonization) for those who underwent decolonization by its efficacy. Infected patients could develop one of the following infections: bacteremia, pneumonia, tracheobronchitis, airway infection (e.g., endobronchial lesions), or other infections (empyema, pericarditis, peritonitis, mediastinitis, and non-superficial SSIs) [2]. All patients in the model could experience graft rejection and all-cause mortality, regardless of colonization and infection status.

Fig. 1
figure 1

General model structure (top) and S. aureus infection outcomes subtree (bottom). The model and infection outcomes are described in the “Methods” section

Full size image

MSSA and MRSA infections were treated with intravenous (IV) nafcillin (2 g every 4 h) and IV vancomycin (1 g, twice a day), respectively; serum vancomycin troughs were measured two to three times. Tracheobronchitis, airway infections or other infections were treated for 14 days. Pneumonia was treated for 3 weeks, and a thoracic computed tomography (CT) scan was performed. Bacteremia was treated for 4 weeks and blood cultures were performed twice. A peripherally inserted central catheter (PICC) line was needed for drug treatment. All patients with known colonization and/or infection were placed under contact isolation precautions (i.e., use of gloves and gown for each patient contact) for the duration of their length of stay (LOS).

Tables 1 and 2 display the model input parameters with their corresponding values and sources. All values came from UPMC experience or published literature; expert opinion was employed if no data were available. The median patient age was 59 years (range, 16–81 years); where applicable, inputs were age-dependent. Our model determined the economic value of screening by the incremental cost-effectiveness ratio, calculated as:

$$ {\mathrm{Cost}}_{\mathrm{screening}}\hbox{--} {\mathrm{Cost}}_{\mathrm{no}\ \mathrm{screening}}/{\mathrm{Effectiveness}}_{\mathrm{screening}}\hbox{--} {\mathrm{Effectiveness}}_{\mathrm{no}\ \mathrm{screening}} $$

This was calculated for each simulation, where effectiveness was measured in quality-adjusted life years (QALYs). Our model assumed a 5-year life expectancy for all patients post-transplant, based on a 50 % mortality rate at 5 years [12]. Patients in the model received age and transplant-adjusted QALY values (discounted to the present year) for all 5 years, unless death occurred. Patients received an in-hospital QALY value for their in-hospital time post-transplant, a 1 year post-transplant QALY value for the remainder of the year, and an over 2 year post-transplant QALY value for another 4 years (assuming a 5-year life expectancy) unless death occurred (Table 1). Patients with a S. aureus infection received an adjusted in-hospital QALY value, based on their specific condition. Effectiveness was also measured in S. aureus infections averted by screening. Economic value from the hospital perspective was measured in lost bed days, following a method described by Graves that is the established standard in economic analyses [13]. A patient with a S. aureus infection experienced an increased LOS; therefore, the patient occupies a bed that in the absence of the infection could be used by another patient. As such, this situation represents an opportunity cost (i.e., lost revenue) to the hospital. The third party payer perspective included all direct costs of an infection, including hospitalization, diagnosis, and treatment costs. All costs were converted to 2012 $US using a 3 % discount rate.

Table 1 Model cost and utility inputs
Full size table
Table 2 UPMC clinical experience outcomes parameter data
Full size table

Simulations and sensitivity analysis

Each simulation sent 1,000 lung and heart-lung transplant recipients through the model 1,000 times for a total of one million trials with unique outcomes. In each simulation, probabilistic sensitivity analysis varied each parameter throughout the ranges listed in Tables 1 and 2 independently, unless explicitly linked in the model. In addition to analyzing the experience at UPMC, separate sensitivity analysis looked at the effect of varying S. aureus colonization (5–15 %), probability of infection (10–30 %), and decolonization efficacy (25–90 %).

Results

UPMC baseline rates

The outcome and LOS data based on the 5-year clinical experience among 596 lung and heart-lung transplant recipients at UPMC are presented in Table 2. The UPMC baseline rate of S. aureus colonization was 9.6 %. Among colonized patients, the baseline rate of post-transplant S. aureus infections in the absence of decolonization was 36.7 %. S. aureus infections were diagnosed in 25 % of colonized patients after the recommendation for a decolonization protocol was introduced, resulting in a theoretical decolonization efficacy of 31.9 %. In contrast, the rate of S. aureus infection among non-colonized patients was 16.3 %.

Costs per S. aureus case prevented

Under all scenarios tested in our model, S. aureus screening and decolonization was economically dominant over no screening or decolonization (i.e., less costly and more effective). The cost savings per case averted specifically at UPMC were $240,602 from the hospital perspective. The cost per case averted for various rates of colonization, infection and decolonization efficacy is presented in Table 3, where negative values denote savings. As shown, the savings from the hospital perspective ranged from $73,567 to $133,157. From the third party payer perspective, savings ranged from $10,748 to $16,723. The trend in cost savings per case averted was similar across perspectives: as the probabilities of infection and colonization increased, the cost savings per case averted increased (Table 3). A similar trend was seen for decolonization efficacy, although to a lesser degree.

Table 3 Cost per S. aureus infection averted using screening and decolonization at various S. aureus colonization and infection probabilities
Full size table

S. aureus infections averted

Figure 2 shows the number of cases averted for the various scenarios. Using UPMC clinical data, 11.3 S. aureus infections (7.3 MRSA infections and 4.0 MSSA infections) were prevented per 1000 patients screened, and 89 lung transplant recipients needed to be screened and decolonized to prevent one S. aureus infection. Therefore, 6.7 S. aureus infections (4.3 MRSA and 2.4 MSSA) could have been averted among the 596 UPMC patients.

Fig. 2
figure 2

S. aureus infections averted at various rates of colonization, probability of infection and decolonization efficacy. Data are presented as the numbers of infections averted per 1000 patients screened

Full size image

As evident in Fig. 2, the number of cases averted varied by the colonization rate. S. aureus screening and decolonization averted fewer cases when the colonization rate was 5 % than when the rates were higher. Figure 2 shows that the rate of colonization and probability of infection had a greater effect on the number of cases averted than decolonization efficacy. Especially at low colonization rates, the efficacy of the decolonization had little effect on the number of cases averted. As the probability of colonization increased, the gains by increasing decolonization efficacy became more prominent by averting more cases.

Discussion

Computer simulation modeling indicates that screening and decolonization of lung and heart-lung transplant recipients can prevent S. aureus infections while saving costs. The data suggest that 6.7 S. aureus infections could have been prevented in our program with routine screening and decolonization over the study period. Based on our expected costs per case averted, the total savings to our institution over 5 years were estimated to exceed $1.6 million. Although our outcomes parameter estimates were derived from UPMC data, the results highlight a potential economic benefit for other programs. Indeed, the benefits of screening and decolonization in our model were evident at low rates of S. aureus colonization (5 %), infection (10 %) and decolonization efficacy (25 %). Screening and decolonization were even more cost-effective at higher rates of S. aureus colonization and infection. As our S. aureus rates are lower than what can be seen in surgical populations (e.g., 40–60 % for general surgery patients), the cost savings per infection averted would be substantial at higher colonization rates (exceeding $133,157 with a colonization rate >15 %, hospital perspective). As such, lung transplant programs can use the infection and decolonization efficacy rates as benchmarks to estimate their anticipated savings per case averted, and compare our analyses with local epidemiology to decide if implementing routine screening is an economically valuable option.

The economic value of S. aureus screening and decolonization among lung and heart-lung transplant recipients was greater than in previous models of orthopedic, vascular and cardiac surgery patients, which also revealed substantial benefits [8–10]. In part, our results likely reflect the high costs of caring for lung and heart-lung transplant recipients in general, especially at programs like ours with high percentages of older and high-risk recipients [1, 14]. It is plausible that any post-transplant complication that extends LOS or requires further medical interventions following lung transplantation would have a particularly strong economic impact. Indeed, we demonstrated previously that S. aureus infections after lung and heart-lung transplantation were associated with prolonged stays in the hospital (median, 43 days with S. aureus infection vs. 24.5 days without infection) and ICU (median, 14.5 days vs. 5 days), and long-term sequelae like increased acute cellular rejection, chronic rejection, and 1- and 3-year mortality [1]. Therefore, our experience predicts that successful decolonization strategies would yield both direct and indirect benefits from reducing S. aureus infections. Taken together, the data suggest that screening and decolonization strategies directed at lung and heart-lung transplant recipients may be more efficient than other high-risk populations. A randomized clinical trial to conclusively demonstrate the benefit of such a strategy is warranted.

A limitation of our study is that the decolonization protocol was not mandatory during the study period, and we do not have data on compliance within the CT-ICU. For the purposes of modeling, we used the 2010 clinical data as representative of our experience after the protocol was introduced. However, we cannot draw definitive conclusions about the clinical impact of screening and decolonization among our patient population. Nevertheless, the data were within reasonable expectations for what might be achieved. Moreover, screening and decolonization were effective under all scenarios tested, over a wide range of colonization, infection and decolonization efficiency rates. Of course, it must be acknowledged that computer models are simplifications of real life and cannot represent all situations and outcomes that may result from S. aureus colonization or infection [15]. In addition, our model utilized colonization and infection rates from a single center, and included S. aureus infections that occurred within 90 days of lung transplantation. As these rates were lower in our population than in other surgical populations, our results are conservative about the economic value of screening and decolonization. We believe the 90-day timeframe is appropriate, since it represents the period of peak immunosuppression and infection risk. The model cannot fully reflect the sociodemographic and clinical differences that may be present among patients across institutions.

In conclusion, our data demonstrate that S. aureus screening and decolonization of lung and heart-lung transplant recipients is economically dominant, capable of both saving significant costs and providing health benefits. Future studies should explore different decolonization strategies or determine the economic advantage of screening lung and heart-lung transplant patients for other pathogens.