This is a decision analytic model comparing the costs and effects of the Pleurx® catheter to chest tube placement with talc pleurodesis for treatment of malignant pleural effusion. describes base-case model inputs. Details of the model and data are described below.
This analysis includes patients over 50 years of age with recurrent MPE with a goal to achieve successful pleurodesis and relief of symptoms. Patients with any type of cancer and MPE are eligible for inclusion. This analysis pertains only to the treatment for a single effusion. We assume that the diagnosis of MPE has already been established at the time of the intervention.
Perspective and time horizon
Analysis is performed from the perspective of the third-party payer; only direct health care costs to patients and direct insurance-covered costs are considered. We did not include direct non-health care costs (such as costs of transportation and child care) and productivity costs in our analysis. The time horizon for this model is 6 months.
This analysis was performed using a static decision tree model. The decision tree is shown in , and was constructed using Tree Age 4.0 software (Treeage Software Inc., Williamstown, MA, 2009). Analysis was performed as a cost-effective analysis, with costs valued in 2008 dollars and effects valued in utilities (which were transformed to quality-adjusted life years, QALYs). The decision node (square shape, MPE) in the tree represents the alternate treatment approaches of Pleurx® catheter placement versus talc pleurodesis for treatment of MPE. Only serious complications of infection and death are considered. These complications are represented as branch points from the chance node (circle shape) entitled “complications” and “no complications” after each treatment option.
Decision Tree: Pleurx Catheter versus Talc Pleurodesis.
Other less common complications, such as fever and pain, are not included; we assume these complications would be managed conservatively, do not incur differential additional costs, and effects on utilities are transient.
Effects data: outcomes and assumptions
The studies that we reviewed refer to treatment success with talc or Pleurx® placement as resolution of the effusion and catheter removal. Thus, our treatment outcome (represented by terminal nodes [triangles]) is effusion resolution (, “eff resolved” and “eff not resolved”). If the effusion is resolved, symptoms related to the effusion are considered to have resolved. We assume that patients still live for the remainder of the 6 months with a baseline level of fatigue even if the effusion has resolved. We also assume that these treatments are palliative; they have no differential effects on mortality and all effects are expressed in terms of patient utilities.
Effects data: treatment success/time to success
Base-case and range estimates for treatment effects are reported in , with a more detailed explanation of these estimates shown in and . Estimates for success of treatment with talc ranged from 0.62 to 0.80. Based on a recent large meta-analysis, we chose 0.80 as the base-case estimate for success of talc pleurodesis.12
Several studies were identified for treatment with Pleurx®
(), and from these studies, we defined the base-case probability of success with the catheter as 0.45. This value was chosen after careful review of the literature with more weight placed on those studies with larger numbers of patients and higher methodological quality.
Using data shown in , we chose the “time to pleurodesis” for patients receiving Pleurx® catheter to be 10 weeks. This time point represents the most clinically relevant estimate based on our consultation with a local thoracic surgeon. This variable was important in the costing of supplies for those with the catheter, and for calculating utilities for those with resolved or unresolved effusions.
Effects data: treatment failure
The clinical approach for those who fail initial treatment for MPE is not well defined in the literature. We assume that patients who do not respond to talc treatment rarely receive a second talc treatment, and would be managed conservatively with medications and/or oxygen. Thus, we do not include an option for further treatment for those for whom treatment with talc fails. With regard to those treated with the catheter, a small minority of patients who fail to obtain effusion resolution elect talc treatment for symptom control.
We therefore performed a separate analysis and included a branch for second-line talc treatment only in those with Pleurx® placement without effusion resolution. We assume that 10% of patients will elect this second-line treatment. We assume that those with complications after Pleurx® treatment will not elect to have an additional chest tube placement and talc procedure.
Effects data: treatment complications
and document treatment complication rates. The most frequently reported complication for both procedures was infection. The most infrequently reported complication was death, but given the serious nature of this complication, it was included in the model. Other reported complications were fever and pain, but for model simplicity we assumed that these complications did not incur additional costs or change utility estimates.
The base-case probability of infection with talc was defined as 0.01 with a range of 0.01 to 0.05. The base-case probability of death with talc was 0.005, with a range of 0.005 to 0.03. To obtain an overall probability of complications for talc, we summed the base-case probabilities to obtain an overall probability of complications of 0.015. We then adjusted the base-case probabilities to reflect conditional probability, and calculated a probability of infection (given complications) as 0.667 and probability of death (given complications) as 0.333 ().
The main complication for catheter treatment was infection, and reports of infection probabilities in the literature () ranged from 0.02 to 0.13. When considering data from studies with the most patients, and using more recently published studies, we chose the base case probability of infection with Pleurx® as 0.075. No studies reported on deaths from the catheter, but since this outcome was viewed as clinically possible and relevant, we estimated this probability to be 0.00001. We defined the probability of complications as 0.07501, and calculated a probability of infection given complications as 0.9999 and a probability of death given complications as 0.0001 ().
An important assumption of our model is that “infection” from either talc treatment or Pleurx® catheter is equivalent to an infection of the pleural fluid (empyema). We also assume that an infection requires 1 week of hospitalization and placement of a chest tube.
The utilities for patients with pleural effusion were obtained from a recent study documenting health state utilities for lung cancer.13
In this study, the authors document societal utility values (with a utility of 1.0 defined as perfect health, and utility of 0 defined as dead) for patients with cancer. We used the utilities reported for the following states in our analysis: “progressive cancer” (0.473) corresponds to those with an unresolved effusion and continued dyspnea, and “cancer, responding and fatigue” (0.599) corresponds to those with a resolved effusion. No utility was given for hospitalization, so we estimated this utility as 0.40 ().
To assign a utility for patients with the catheter, we used the results of a recent meta-analysis of patients on peritoneal dialysis (PD).14
We chose the peritoneal dialysis population because both patient groups (those with Pleurx®
and those on PD) live with an indwelling catheter that is accessed regularly for drainage. From this analysis, we inferred a base-case utility for patients with Pleurx®
catheter as 0.58.
Cost data was estimated using Medicare Diagnosis-Related Group (DRG) reimbursement data from 2008. Cost data is summarized in : the DRG for “pleural effusion without complications” (DRG: 188) was used to cost admission for talc and the DRG for “pleural effusion with complications” (DRG: 186) was used to cost admission for infection. The Medicare Current Procedural Terminology (CPT) code for placement of the pleural catheter was used for this cost variable, and the cost estimate for Pleurx® supplies was obtained from several Internet vendors and verified with a local apothecary.
The cost of a nursing visit was obtained from a cost-effectiveness study of pediatric home visits15
and was adjusted for inflation to reflect 2008 dollars (). This cost was also verified with a local home nursing agency. The cost of a physician office visit was estimated from billing data at our institution. In both treatment arms, patients require one monthly visit with a physician. We also assume that patients with Pleurx®
require three visits from a nurse per week. In all patients for whom the effusion does not resolve, we assume that home nursing visits occur once per week for the remaining month(s) of life.
We determined the incremental cost-effectiveness ratio (ICER) of Pleurx® catheter placement compared with talc pleurodesis for the base-case assumptions. In circumstances where the catheter was less effective and more costly than talc pleurodesis, we considered the catheter to be dominated. When the catheter is both more effective and more costly compared with the talc pleurodesis we calculate and report an incremental cost-effectiveness (C/E) ratio. For incremental ratios up to $100,000/QALY, we consider such ranges as likely to be cost effective.
To examine the stability of our results, univariate sensitivity analysis was performed. For each analysis, one parameter of the model was varied over a range as defined in , while keeping other parameters at their baseline. Parameters considered for sensitivity analysis were those that appeared most variable in the literature review.