TreeAge Pro includes many analysis features. Several of the most commonly used features are described on this page.
Calculate the expected value (EV) at every node in the model.
Click to Expand Image
In the example above, you see the payoff expressions evaluated at the terminal nodes to generate the EV for each possible scenario. The scenario EVs then feed back to the chance nodes by calculating a weighted average based on each branch's EV and probability. The EVs for the two strategies are then compared based on a willingness-to-pay parameter ($50K) set in the Tree Preferences, and the most cost-effective strategy is selected as optimal.
Compare all strategies in a cost-effectiveness tree by presenting them in a Cost-Effectiveness Graph and by calculate all Incremental Cost-Effectiveness Ratios (ICERs).
Click to Expand Image
The CEA graph above shows the two strategies based on cost and effectiveness. The line connecting the two strategies is the cost-effectiveness frontier. Dominated strategies (absolute or extended) would not be on the cost-effectiveness frontier and would be rejected. The slope of the line is the incremental cost-effectiveness ratio (ICER).
The CE rankings report above shows the calculation of incremental cost, incremental effectiveness and ICER. If the willingness-to-pay threshold were greater than the ICER, the more expensive and more effective strategy would be recommended.
Show the accumulation of cost and/or effectiveness in a Markov model on a cycle by cycle and state by state basis.
Click to Expand Image
The Markov Cohort output above demonstrates the movement of the cohort among the states by cycle. The transitions in the output follow the transitions in the model, allowing you verify state progressions with time.
Evaluate how specific uncertainties (via variables) affect EV and CEA calculations. Then generate text and graphical output to illustrate the effects of uncertainty and value thresholds.
Click to Expand Image
The sensitivity analysis output above shows EV and ICER calculations for a range of values for a specific variable (pEradicateRadSurg). You can see the ICER decrease as the variable value increases. With a WTP threshold of $50K, the variable threshold is between 0.725 and 0.75.
Click to Expand Image
The graph above shows the ICER decreasing as the variable value increases. You could add horizontal and vertical lines at the WTP threshold to try to find the variable threshold, but the threshold is easier to see in the graph below.
Click to Expand Image
The graph above uses Net Monetary Benefit (NMB) to combine cost, effectiveness and WTP into a single value. The strategy with the highest NMB is the most cost-effective given the fixed WTP. This makes it easy to see the threshold value for the variable at 0.749.
Evaluate how parameter uncertainty affects EV calculations when any number of parameters is sampled from distributions. Then generate text and graphical output to highlight how uncertainty affects confidence in strategy recommendations.
Click to Expand Image
The output above shows the aggregates statistics for all the PSA simulation iterations as well as all the text and graph options associated with the simulation.
Click to Expand Image
The ICE scatterplot graph plots each iteration based on its incremental cost and incremental effectiveness. The dashed line represents the WTP threshold of $50K. All iterations below and to the right of the WTP line represent an ICER below $50K, so the more costly, more expensive treatment option is considered cost-effective.
Click to Expand Image
The Acceptability Curve shows the percentage of PSA iterations that find one or the other treatment option optimal given different WTP values. As the WTP increases, more and more of the iterations favor the more costly, more effective treatment option.
Run individual random walks (trials) through the model to allow for heterogeneous cohorts and the tracking of events. The mean values from the individual patient histories become EV estimates that can be used for cost-effectiveness analysis.
Click to Expand Image
The output above shows the aggregates statistics for all the individual trials that ran through the model as well as all the text and graph options associated with the simulation. The mean values (cost and effectiveness) for each strategy are estimates of expected value. A sufficient number of trials must be used, so that the EV estimate is stable and accurate.
Click to Expand Image
The text report above shows individual trial data for each strategy. The COST and EFF columns reflect the final patient outcomes. The T_STROKE... columns reflect the number of stroke events. The DIST_1 column reflects the starting age for each individual trial - the same value is used for all strategies.
Click to Expand Image
The CEA graph above (rankings available too) is generated from the mean values (EV estimates) generated from the Microsimulation.
|