BACKGROUND. The optimal management of pediatric empyema is controversial. The purpose of this decision analysis was to assess the relative merits in terms of costs and clinical outcomes associated with competing treatment strategies.
METHODS. A cost-effectiveness analysis was conducted using a Bayesian tree approach. Probability and outcome estimates were derived from the published literature, with preference given to data derived from randomized trials. Costing was based on published estimates from Great Ormond Street Hospital (London, United Kingdom), supplemented by American and Canadian data. Five strategies were evaluated: (1) nonoperative; (2) chest tube insertion; (3) repeated thoracentesis; (4) chest tube insertion with instillation of fibrinolytics; or (5) video-assisted thorascopic surgery. The model was used to project overall costs, survival in life-years, and incremental cost-effectiveness ratios for competing strategies.
RESULTS. In the base-case analysis, chest tube with instillation of fibrinolytics was the least expensive therapy, at $7787 per episode. This strategy was projected to cost less but provide equivalent health benefit when compared with all of the competing strategies except repeated thoracentesis, which had an incremental cost-effectiveness ratio of approximately $6 422 699 per life-year gained relative to chest tube with instillation of fibrinolytics. In univariable and multivariable sensitivity analyses, thorascopic surgery was preferred only when the length of stay associated with chest tube with instillation of fibrinolytics exceeded 10.3 days or when the probability of dying as a result of this strategy exceeded 0.2%, assuming a threshold willingness to pay of $75 000 per life-year gained. Chest tube with instillation of fibrinolytics was preferred in >58% of Monte Carlo simulations.
CONCLUSIONS. On the basis of the best available data, chest tube with instillation of fibrinolytics is the most cost-effective strategy for treating pediatric empyema. Video-assisted thorascopic surgery would be preferred to chest tube with instillation of fibrinolytics if the differential in length of stay between these 2 strategies were proven to be greater than that suggested by currently available data.
Pneumonia is a common cause of pediatric hospitalization, and ≤50% of children hospitalized with pneumonia have an associated parapneumonic effusion.1 Most of these effusions resolve spontaneously with treatment of the underlying pneumonia. However, ∼5% do not respond to antibiotics, usually because of the development of a “complicated” parapneumonic effusion, characterized by a loculated fibropurulent collection, or empyema. There has been a documented increase in the incidence of pediatric empyemas since the mid-1990s in both the United Kingdom2,3 and the United States,4 possibly because of serotype replacement after the universal introduction of heptavalent pneumococcal vaccine in infants3,5 or evolving antibiotic resistance patterns.4,6,7
It is generally accepted that empyemas require additional nonmedical management to effect drainage in the form of repeated ultrasound-guided thoracentesis, thoracostomy (chest tube) drainage with or without instillation of intrapleural fibrinolytics, or drainage using video-assisted thorascopic surgery (VATS) or open thoracotomy techniques. There is considerable controversy regarding the most appropriate therapy. Current published guidelines recommend insertion of an ultrasound-guided small bore percutaneous chest tube (CT) with instillation of fibrinolytics (CTWF) in any patient with a chest drain inserted for an empyema.8 However, proponents of VATS argue that a surgical approach to the primary management of empyema is associated with shorter hospitalization and a reduced risk of treatment failure.9–11 Differences in interpretation of the published literature can be largely attributed to an absence of large, adequately powered, randomized, controlled trials to help inform clinicians.
Although not a substitute for definitive clinical trials, a decision analysis can help guide clinicians in determining the optimal choice of therapy for patients through synthesis of the best available data and by explicitly weighing the expected health and economic benefits of competing strategies. Our objective was to compare the projected costs and clinical outcomes associated with 5 competing strategies for the management of pediatric empyema to determine whether short-term increases in costs associated with new technologies are likely to be counterbalanced by diminished length of stay and improved clinical outcomes.
Decision Analytic Model
We constructed a decision tree (Fig 1) to evaluate the cost-effectiveness of competing strategies for the management of pediatric empyema. We used a Bayesian tree approach to estimate the payoffs of each strategy. The strategies considered were, in order of increasing invasiveness: (1) nonoperative (antibiotics with or without delayed CT insertion); (2) immediate CT insertion without instillation of fibrinolytic agents; (3) repeated thoracentesis; (4) CTWF; or (5) VATS. The model outputs were costs (in US $) and benefits (life expectancy in life-years). Incremental analyses were performed by ranking all 5 of the strategies in order of increasing effectiveness after eliminating strategies that were more costly and less effective than another strategy (simple dominance). We then calculated the incremental cost-effectiveness ratio (ICER) for each strategy, defined as the marginal cost divided by the marginal benefit, compared with the next most expensive option. When a given strategy had a higher cost but lower ICER than the next most effective strategy, the less effective strategy was eliminated by so-called “weak dominance.” All of the analyses were performed on TreeAge Pro 2006 Suite (TreeAge, Williamstown, MA).
Base-case unit cost data expressed in US dollars were obtained from published estimates from a large, internationally recognized pediatric institution, Great Ormond Street Hospital (London, United Kingdom),12 supplemented by recommended rates of reimbursement in Canada or the United States, where available. For sensitivity analyses, we also estimated Canadian costs from our hospital and American costs from published facility and physician charges for empyema management,13 using a cost/charge ratio of 0.4,14 supplemented by the average per-diem Medicare hospital reimbursement for pneumonia and pleurisy in 0- to 17-year-olds,15 physician reimbursement,16 and medication costs.17 All of the costs were converted to 2006 dollars using the medical services component of the consumer price index.18 All of the patients receiving fibrinolytics were assumed to receive 6 doses of urokinase, and costing of 1 to 6 doses of an alternative fibrinolytic agent (tissue plasminogen activator [tPA]) was incorporated into sensitivity analyses. Costs of parenteral antibiotics were not modeled, because they were assumed to be proportionately related to length of stay and could be captured with liberal estimates of the cost per day in hospital. We also did not incorporate outpatient costs after discharge, such as diagnostic imaging and oral antibiotics, because these can be assumed to be identical in all arms of the study as per current treatment guidelines.8 The cost estimates used in the analysis are summarized in Table 1.
Mortality estimates were based on available data. A meta-analysis of observational trials only described mortality in conservatively treated patients.10 In the absence of reports of empyema-related mortality for other therapeutic modalities, we used published data on estimates of attributable mortality from anesthesia (0.005%)19 and lifetime attributable cancer mortality risk from chest computed tomography radiation exposure in a young child (0.100%).20,21 For repeated thoracentesis, we assumed that the procedure was performed on the patient 3 times22 without anesthesia and relied on estimates on the risk of pneumothorax (2.5% per thoracentesis),23 assuming a mortality of 0.05% in patients who developed a pneumothorax. For all of the sensitivity analyses, we examined mortality rates ≤1% for each arm. The base-case costs and probabilities with all of the plausible ranges of variables are presented in Table 2.
The base case was a previously healthy 5-year-old boy with an ultrasound-confirmed empyema and a normal life expectancy of 72.4 years, based on standard life tables available through Statistics Canada (www.statcan.ca/english/freepub/84-537-XIE/tables.htm). An age of 5 years was chosen, because this is a common median age of presentation of empyema in trials, and male gender was chosen, because the condition is more prevalent in boys.24 Base-case probabilities and outcome estimates were derived from the published medical literature, with preference given to grouped means or proportions of pooled data derived from randomized, controlled trials (RCTs) comparing therapeutic strategies; in the absence of RCT data, we used estimates from a meta-analysis of observational studies.10 RCTs were found in PubMed using the search terms “empyema” OR “pleural effusion” AND “randomized, controlled trials” and limiting the search to children (0–18 years). Manual review of the reference lists of identified studies, as well as the reference list of recently published systematic reviews of the pediatric empyema literature,8,10 were also examined to identify other studies. For the repeated thoracentesis arm, trial or pooled data were unavailable, so we used data from the only known published observational study on this therapy in pediatric empyema.22 Liberal estimates of plausible ranges were derived from the 95% confidence interval of the data presented or the next best data source. For instance, if RCT data were used for the base case, then the widest range of data from either the individual RCTs or from the pooled observational data determined the plausible range.
Pediatric trials that described delayed instillation of fibrinolytics after insertion of a CT were analyzed as being in the CT alone strategy. Adult studies were excluded, because it is well accepted that experiences with adults cannot be extrapolated to children because of the high mortality and coexistent illness in adults with empyema.25 All of the children undergoing VATS were assumed to have a single computed tomography chest scan, because most surgeons request this scan be performed before an operation. Other diagnostic imaging scans were excluded from the model, because there are no evidence-based guidelines that recommend the frequency and modality of their use in empyema. All of the patients who failed their primary procedure underwent a salvage procedure. This procedure was either the same or more invasive then the initial therapy. If the salvage procedure was surgical, it was assumed to be VATS. Although some clinicians use an open thoracotomy technique for patients who fail primary therapy, the frequency of this procedure and the costs and effectiveness of this procedure are either unknown or have never been evaluated in a trial. In addition, it was assumed that all of the patients had a maximum of a single salvage procedure. Therefore, if there was no improvement from the salvage procedure, the patient would not require yet another salvage procedure. In the absence of outcome data from trials on salvage procedures, it was also assumed that the cost and outcome of a salvage procedure were the same as those of the primary procedure. Thus, if a patient failed VATS and required salvage VATS, it was assumed that the costs were twice the costs of a single VATS, and the outcome (survival) was the independent product of the outcome of 2 VATS procedures. This is congruent with literature reports of the overall cost of patients needing salvage VATS as approximately double that of patients who underwent primary VATS.9 Outcomes of patients receiving fibrinolytic agents were assumed to be independent of the type of agent used (urokinase or tPA) or the frequency of use (1–6 times per patient).
Because most children with empyema completely recover with no clinically important pulmonary sequelae, such as exercise intolerance,8,26–28 it was also assumed that all of the patients who survived empyema to discharge had a normal quality of life after discharge. In other words, the effectiveness outcome of expected life-years was chosen because it approximates quality-adjusted life-years, the standard metric used in many cost-effectiveness analyses.29
We performed a deterministic 1-way sensitivity analysis on all of the variables in our model to determine the effect of varying baseline estimates within clinically plausible ranges on our results. Variables that were sensitive to changes in baseline estimates were modeled with a 2-way sensitivity analysis.
In addition, we performed Monte Carlo simulations, in which cohorts of simulated patients underwent management of empyema. Such simulations use a random-number generator to create unique, simulated individual patients and move them through a series of chance events over time. A running tally of outcomes, costs, and events is recorded, with the creation of simulated cohorts that can be compared with one another. We performed 1000 second-order simulations, with characteristics of patient cohorts and probabilities, outcomes, and costs drawn from plausible distributions. Each second-order simulation was composed of 1000 first-order probabilistic trials (ie, 1 million trials in total), with parameter values held constant. Probabilities were assumed to follow β distributions, whereas other parameter distributions were assumed to be triangular and bounded by upper and lower bound parameter estimates. Monte Carlo simulations were used to construct cost-acceptability curves,30 with the likelihood that a given strategy would be favored plotted against societal willingness-to-pay for an additional life-year.
The least costly strategy was CTWF with a cost of $7787 and survival of 72.43 years. At a cost of $18 581, repeated thoracentesis had a marginal increase in life expectancy of <1/100th of a year, resulting in an ICER of $6 422 698 per life-year gained relative to CTWF. All of the other strategies had higher cost and lower survival than CTWF. In particular, VATS was $2835 more costly and less effective by 0.07 years than CTWF. The nonoperative arm was associated with a 2.4-year decreased survival and was, thus, the only strategy associated with a marked (>1/10th of a year) difference in survival. Table 2 summarizes the results of the base case for all of the strategies. No qualitative changes in the relative attractiveness of competing strategies occurred with the use of discounted, rather than undiscounted, estimates of life expectancy or with the use of Canadian or US cost estimates.
Model projections were most sensitive to changes in the anticipated length of stay of CTWF. If the baseline length of stay rose beyond a threshold of 10.3 days, VATS became the preferred option. Similarly, assuming a willingness to pay (WTP) of $75 000 per life-year gained, once the probability of mortality from CTWF rose beyond 0.2%, with all of the other strategies remaining equal, VATS became the preferred option. The prediction of CTWF as the preferred strategy was robust to all of the other univariate and bivariate sensitivity analyses. Using a more expensive fibrinolytic agent (tPA) decreased the threshold value where VATS is preferred to a length of stay of 9.2 days. Given that the cost-per-day base-case estimates were conservative and did not incorporate the cost of antibiotics, a 2-way sensitivity analysis of cost per day and length of stay of CTWF is described. As shown in Fig 2, increasing the cost per day up to $5000 does not change the preferred strategy with base-case assumptions but decreases the threshold length of stay, where VATS becomes the preferred strategy. Projections were insensitive to variation in other model parameters across plausible ranges. There were no changes in model projections when we simulated survival in girls rather than in boys.
Monte Carlo Simulation
In probabilistic sensitivity analyses that incorporated parameter distributions and random chance, CTWF was the preferred strategy in >58% of trials, regardless of the societal WTP threshold (Fig 3). The attractiveness of VATS relative to CTWF and other strategies declined with increasing WTP for a life-year, because the cost savings associated with reduced length of stay became less influential. Repeat thoracentesis became a more attractive strategy with increasing WTP, but the likelihood that this strategy was preferred was always <10% for WTP thresholds up to $200 000 per life-year gained.
The preferred strategy in this decision analysis was CTWF. Repeated thoracentesis would be preferred only if societal willingness to pay was more than $6 000 000 per life-year gain, a ratio that is generally not considered cost-effective in most health care settings. The results were robust and only sensitive to changes in either length of stay of CTWF or mortality of CTWF beyond the best available estimates in the literature. CTWF was the preferred strategy, regardless of societal willingness to pay for health, in probabilistic sensitivity analyses.
There have been 4 RCTs,12,13,31,32 1 meta-analysis of observational studies,10 and 1 guideline8 published to date that have compared the therapeutic options presented in this decision analysis. The results of the RCTs are summarized in Table 3. In the 2 trials comparing CT with CTWF, 1 study described a decreased length of stay in patients who received a CTWF compared with a CT with installation of saline,32 whereas another smaller trial31 found no difference in the length of tube insertion in the 2 groups but an increase in the need for surgical drainage in the placebo group. Our model provides additional support for the recommendation that intrapleural fibrinolytic agents decrease the costs associated with treatment of empyema.
Of greater controversy has been the debate over the relative benefit of VATS over CTWF. One small RCT (n = 18) showed a significantly reduced length of stay and a nonsignificant trend toward lower cost in patients who received VATS,13 whereas another larger study (n = 60) demonstrated no difference in length of stay and increased costs with VATS.12 The difference in the results of the 2 studies can be attributed largely to differences in length of stay in the nonsurgical arms (13.2 vs 6.0 days), possibly because of the use of fibrinolytics as salvage rather than primary therapy in the CT arm of the smaller study. A meta-analysis of observational studies10 on this question reported improvement in mortality (0.0% vs 3.3%), reintervention rate (2.5% vs 23.5%), length of stay (10.8 vs 20.0 days), and duration of chest drain insertion (4.4 vs 10.6 days) in operatively treated patients (VATS or thoracotomy) compared with patients treated with fibrinolytic therapy. The frequency of all of these outcomes and the rate of complications with fibrinolytics are much higher in this meta-analysis than those reported in the aforementioned RCTs. This could be partially attributed to limitations in the meta-analysis resulting from the inclusion of observational and largely retrospective studies with heterogeneous patient populations from disparate parts of the world. For instance, some studies included children with simple effusions, and there are many potential confounders for all of the outcomes presented. Critics of this review have argued that the systematic reviews of this topic should preferentially use the best available data from published RCTs as opposed to simply pooling observational studies.33 Our decision analysis incorporated this recommendation to derive base-case estimates from randomized, controlled trials whenever possible, although we did use other data sources for sensitivity analyses. Our projections were robust in the face of variation of model parameters related to costing and failure rates.
The major limitation in constructing this model is the inconsistent and limited published data available on this topic. Few RCTs have been published, and all are small, with a maximum of 60 patients enrolled. Attempts to synthesize data from observational trials for pediatric empyema are limited by the varying quality of these largely retrospective studies that make the pooling of results problematic.33 In addition, most surgical trials are contemporary, whereas many nonsurgical trials are not, making time an important source of bias in the assessment of outcomes.9 There is marked variation in the timing of interventions for patients in the published literature. It is well recognized that “late” interventions can be associated with prolonged recovery in both medical34 and surgical7 therapies.
Another limitation in this model is the use of length of stay as an important predictor of cost. Although length of stay has some face validity in that it is affected by the course of diseases and treatments, it is also influenced by a number of other extraneous factors, such as patient characteristics, physician preferences, and hospital policy. Also, important attributes of suboptimal health states (eg, pain associated with recovery from a procedure) were not captured at all in this model. Length of stay does serve as a proxy for this outcome, because most patients in the hospital for prolonged periods suffer similar pain, as they tend to all have CTs in situ. Also, given that empyema is an acute condition of relatively short duration, it is not anticipated that estimates of disuse of health states would have changed the results of the cost-effectiveness analysis. Parental preferences are also important to consider but were not modeled, because these were not readily available and were beyond the scope of this decision analysis.
Most of the cost estimates chosen for use in the study were obtained from published data from a major pediatric hospital and replicated with North American data. However, the unit costs may not be applicable to all health settings. For instance, the estimated cost of the fibrinolytic agent used in the United Kingdom (urokinase) is different from published costs of other fibrinolytic agents (eg, tPA). Costing of tPA varies markedly depending on dosing protocol and location, ranging from as low as $90 per child in our institution, where tPA has been used successfully in a single dose,34 to as high as $1637 per child for a protocol of 6 doses of tPA using American costing estimates.17,18 Urokinase is currently not available in the United States. Although more study is needed, the best available data seems to suggest that the efficacy of tPA and urokinase in pediatric empyema is similar.35
Despite the limitations outlined, this decision analysis can guide clinical decision-making and inform the rational allocation of health resources. In particular, for proponents of VATS, the potential benefits of cost savings from small decreases in length of stay for VATS are offset by the difference in upfront costs and potentially the hypothetical risk associated with exposure to ionizing radiation from a CT scan. Future trials intended to test the hypothesis that VATS is superior over CTWF need to be designed and powered to show more than just a difference in length of stay, but either a difference in length of stay that is substantial enough to justify the additional expenditures and risks associated with this procedure or evidence for substantial advantages in efficacy as determined by improvements in long-term lung function or exercise tolerance with VATS.
Based on commonly accepted cost-effectiveness thresholds and the best available data, CTWF is the most cost-effective strategy for treating pediatric empyema. Currently, in most centers, the “preferred” treatment is realistically driven by local expertise (eg, a surgeon who can perform VATS, or an interventional radiologist who can insert an image-guided flexible percutaneous catheter), as well as health care provider and caregiver preferences. However, as a guide to the allocation of resources, this cost-effectiveness analysis does support the contention that VATS is not superior to CTWF. VATS would be preferred to CTWF only if the differential in length of stay or survival between these 2 strategies was proven to be greater than that suggested by currently available data.
We thank Drs Ahmed Bayoumi and Sanjay Mahant for helpful comments and Drs Yvonne Yau and Karen Thomas for advice on model estimates.
- Accepted October 19, 2007.
- Address correspondence to Eyal Cohen, MD, MSc, 555 University Ave, Toronto, Ontario, Canada M5G 1X8. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
The optimal management of empyema and complicated parapneumonic effusion is controversial. Recent trials and reviews have produced conflicting results, particularly on the relative merits of a CTWF compared with VATS.
What This Study Adds
CTWF is the most cost-effective strategy for treating empyema. VATS would be preferred to CTWF if the differential in length of stay between these 2 strategies were greater than that suggested by published data.
- ↵Lichtenstein R, Suggs A, Campbell J. Pediatric pneumonia. Emerg Clin N Am.2003;21 (2):451– 473
- ↵Gupta R, Crowley S. Increasing paediatric empyema admissions. Thorax.2006;61 (2):179– 181
- ↵Spencer D, Iqbal S, Hasan A, Hamilton L. Empyema thoracis is still increasing in UK children. BMJ.2006;332 (7553):1333
- ↵Byington C, Spencer L, Johnson T, et al. An epidemiological investigation of a sustained high rate of pediatric parapneumonic empyema: risk factors and microbiological associations. Clin Infect Dis.2002;34 (4):434– 440
- ↵Hardie W, Bokulic R, Garcia V, Reising S, Christie C. Pneumococcal pleural empyemas in children. Clin Infect Dis.1996;22 (6):1057– 1063
- ↵Schultz K, Fan L, Pinsky J, et al. The changing face of pleural empyemas in children: epidemiology and management. Pediatrics.2004;113 (6):1735– 1740
- ↵Balfour-Lynn I, Abrahamson E, Cohen G, et al. BTS guidelines for the management of pleural infection in children. Thorax.2005;60 (suppl 1):i1– i21
- ↵Avansino J, Goldman B, Sawin R, Flum D. Primary operative versus nonoperative therapy for pediatric empyema: a meta-analysis. Pediatrics.2005;115 (6):1652– 1659
- ↵Kurt B, Winterhalter K, Connors R, Betz B, Winters J. Therapy of parapneumonic effusions in children: video-assisted thorascopic surgery versus conventional thoracostomy drainage. Pediatrics.2006;118 (3). Available at: www.pediatrics.org/cgi/content/full/118/3/e547
- ↵US Department of Health and Human Services, Centers for Medicare and Medicaid Services. Hospital inpatient prospective payment systems and 2005 FRY rates. Fed Regist.2004;69 (96):28748– 28749
- ↵US Department of Health and Human Services, Centers for Medicare and Medicaid Services. MEDPAR limited data set (LDS). Available at: www.cms.hhs.gov/MedicareFeeforSvcPartsAB/03_MEDPAR.asp. Accessed August 23, 2007
- ↵Gallagher P. Medicare RBVRS: The Physician's Guide 2002. Chicago, IL: American Medical Association; 2002
- ↵Cohen H. Drug Topics Red Book. Montvale, NJ: Thompson PDR; 2003
- ↵Bureau of Labor Statistics. Most requested statistics: consumer price index. Available at: http://data.bls.gov/cgi-bin/surveymost?cu. Accessed August 18, 2007
- ↵Brenner D, Elliston C, Hall E, Berdon W. Estimated risks of radiation-induced fatal cancer from pediatric CT. AJR Am J Roentgenol.2001;176 (2):278– 296
- ↵McLaughlin F, Goldmann D, Rosenbaum D, Harris G, Schuster S, Strieder D. Empyema in children: clinical course and long-term follow-up. Pediatrics.1984;73 (5):587– 593
- ↵Thomson A, Hull J, Kumar M, Wallis C, Lynn IB. Randomised trial of intrapleural urokinase in the treatment of childhood empyema. Thorax.2002;57 (4):343– 347
- ↵Bollen C, Hoekstra M, Arets H. Pooling of studies in meta-analysis of observational research leads to spurious results. Pediatrics.2006;117 (1):261– 262
- ↵Weinstein M, Restrepo R, Chait PG, Connolly B, Temple M, Macarthur C. Effectiveness and safety of tissue plasminogen activator in the management of complicated parapneumonic effusions. Pediatrics.2004;113 (3). Available at: www.pediatrics.org/cgi/content/full/113/3/e182
- ↵Callum JL, Pinkerton PH, eds. Bloody Easy: Blood Transfusions, Blood Alternatives and Transfusion Reactions—A Guide to Transfusion Medicine. Toronto, Canada: Sunnybrook and Women's College Health Sciences Centre; 2004
- Copyright © 2008 by the American Academy of Pediatrics