The Cost Effectiveness of Implantable Cardioverter-Defibrillators
Results From the Multicenter Automatic Defibrillator Implantation Trial (MADIT)-II
Jack Zwanziger, PhD*,*,
W. Jackson Hall, PhD ,
Andrew W. Dick, PhD*,
Hongwei Zhao, DSc,
Alvin I. Mushlin, MD, ScM*,
Rebecca Marron Hahn, MPH*,
Hongkun Wang, MA,
Mark L. Andrews, BBS ,
Cathleen Mooney, MS*,
Hongyue Wang, MA and
Arthur J. Moss, MD
* Department of Community and Preventive Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York
Department of Biostatistics and Computational Biology, University of Rochester School of Medicine and Dentistry, Rochester, New York
Department of Cardiology Division of the Department of Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York
View larger version (17K):
[in a new window]
|
Figure 1 Incremental cost-effectiveness ratios (iCERs) for various horizons, actual and projected. The iCER was estimated from the data with horizons of 1, 1.5, 2, 2.5, 3, and 3.5 years, marked by circles. The vertical line segment is a 95% confidence interval for the iCER at 3.5 years (Table 3). The other lines are iCERs derived from projected survival curves (this figure) and the regression model for costs (Fig. 2). As in Figure 2, the actual iCER values are slightly higher than the fitted model projections because early replacement costs of implantable cardioverter-defibrillators were removed when fitting the model, with all replacement costs imposed at 5 and 10 years. Note that the vertical axis has a logarithmic scale.
|
|
View larger version (18K):
[in a new window]
|
Figure 2 Survival curves, actual and projected. For the first 3.5 years, the two curves are the Kaplan-Meier estimates of the probability of survival in the implantable cardioverter-defibrillator (ICD) and conventional-therapy (CONV) arms, based on 1,095 patients in this cost study. From 3.5 years onward, the CONV survival curve has been projected assuming proportional hazards with an age- and gender-matched U.S. subpopulation, with a hazard ratio of 4.54. There are three projected ICD survival curves, ICD1 with hazard rates relative to CONV equal to 0.677 throughout, ICD2 with this hazard ratio increasing linearly to unity at 12 years, and ICD3 with the HR increasing to 1.44 at 12 years, reaching 1.0 at 7.1 years; see the text for rationales for these choices. Years-of-life saved are the areas between ICD and CONV curves.
|
|
View larger version (18K):
[in a new window]
|
Figure 3 Cumulative discounted costs, actual and projected. Actual cumulative discounted costs are plotted for the implantable cardioverter-defibrillator (ICD) and conventional-therapy (CONV) arms up to 3.5 years. Projected costs are based on a regression model and plotted over the range from 0 to 12 years; up to 3.5 years, they use the regression model for costs together with Kaplan-Meier estimates of survival, whereas after 3.5 years, there are three variations for the ICD arm projections, corresponding to three assumptions about projected survival, and one CONV arm projection (Fig. 1 legend). The actual ICD arm costs are slightly higher than the projected values because, in projections, the costs of 32 replacement ICDs that occurred in the observation period were removed (to fit a regression model), and all replacement costs were assumed to occur at exactly 5 and 10 years, resulting in jumps in the ICD arm curves at those times. The jump near the origin is caused by the initial ICD implantation.
|
|
|
