A comparison of treatment strategies for hypoplastic left heart syndrome using decision analysis
Pamela C. Jenkins, MD, PhD*,
Michael F. Flanagan, MD, FACC*,
James D. Sargent, MD*,
Charles E. Canter, MD, FACC ,
Richard E. Chinnock, MD ,
Kathy J. Jenkins, MD, MPH ,
Robert N. Vincent, MD, FACC||,
Gerald T. O’Connor, PhD, DSc, FACC¶ and
Anna N. A. Tosteson, ScD¶
* Department of Pediatrics, Dartmouth Medical School, Hanover, New Hampshire, USA
Division of Pediatric Cardiology, St. Louis Children’s Hospital, St. Louis, Missouri, USA
Department of Pediatrics, Loma Linda University Children’s Hospital, Loma Linda, California, USA
Department of Cardiology, Children’s Hospital, Boston, Massachusetts, USA
|| Department of Pediatric Cardiology, Emory University School of Medicine, Atlanta, Georgia, USA
¶ Department of Medicine and Community and Family Medicine and the Center for the Evaluative Clinical Sciences, Dartmouth Medical School, Hanover, New Hampshire, USA
View larger version (16K):
[in a new window]
|
Figure 1 The decision tree comparing surgical treatment strategies for hypoplastic left heart syndrome (HLHS) is shown. Once HLHS is diagnosed, a decision (open square) is made as to which treatment strategy to undertake. After the strategy is decided, events (open circles) occur, with defined probabilities for mutually exclusive outcomes, "survive" or "die." Survival to surgery or to listing for transplantation (Tx) is not certain, and a probability is associated with that survival. All babies surviving to surgery are assumed to receive that treatment as planned. The outcome measure was survival to one year.
|
|
View larger version (21K):
[in a new window]
|
Figure 2 The probabilities of receiving a transplant and of waiting-list mortality, depicted on the vertical axis, are shown as a function of age in days on the horizontal axis. Both probabilities increase with time on the waiting list. Data are derived from the hypoplastic left heart syndrome dataset (3).
|
|
View larger version (34K):
[in a new window]
|
Figure 3 Threshold values of one-way sensitivity analyses, which vary each probability separately, are shown. A threshold value is the value of the probability at which the optimal strategy changed. Thresholds were obtained in the probabilities (shown along the vertical axis) for stage 1 mortality, stage 2 mortality, organ availability and mortality after transplantation. Therefore, these probabilities are most important to the decision. The optimal strategy for particular values of each probability is shown by the pattern corresponding to the legend (right). For example, a center with a combined stage 1 mortality of 15% would optimize survival by offering staged surgery, whereas a center with a stage 1 mortality of 30% would optimize survival by listing the patient for one month before performing stage 1 surgery if no donor is found. No thresholds were found for other probabilities. Black dots indicate the baseline values in the decision tree. The baseline values all lie in probability ranges that favor listing the patient for one month. Tx = transplantation.
|
|
View larger version (103K):
[in a new window]
|
Figure 4 Two-way sensitivity analysis shows how the optimal strategy changes as the probability of organ donation in three months (horizontal axis) and the probability of combined stage 1 mortality (vertical axis) are varied. All other probabilities are retained at baseline, as in Table 2. Strategies producing optimal survival at different values of organ availability and stage 1 mortality were: 1) Staged Surgery; 2) list for one month then perform stage 1 surgery if no donor is found (List 1 Month); 3) list for three months then perform stage 1 surgery if no donor is found (List 3 Months); and 4) list indefinitely until a donor is found (List to Tx). For example, if stage 1 mortality was <20%, staged surgery was the optimal choice. With a moderate to high organ availability (more than 30% in three months) listing for transplantation (Tx) for one month or more would provide the highest survival. The star indicates the decision-tree baseline probabilities of organ donation in three months, 0.64; and stage 1 mortality, 0.48.
|
|
|
