Transitions is the fifth stop on the hēRo3 subway. On this step, you may define how patients move through the health states included in the model. The transitions page functions differently dependent on what type of model structure is used.
Markov Cohort Model
In a Markov Cohort Model, the transitions page defines the transition matrices for each strategy. This is done using a longform table structure, in which each row represents a cell in the transition matrices in one or all strategies, and with the following columns:
|
Column |
Description |
|
Strategy |
The strategy for which the transition being defined applies. To make a transition apply to all strategies, select ‘All’. |
|
From |
The state from which patients transition. |
|
To |
The state to which patients transition. |
|
Formula |
A formula which will calculate the probability of undergoing a given transition. This formula may reference model time or state time in order to introduce time-dependency or the creation of tunnel states, respectively. The C keyword can be used to automatically calculate the complementary probability (i.e. one minus the probability of going to all other states). |
Please note that for the transitions page to be valid for a Markov Cohort model, there must be at least one transition defined for each health state and the sum of all transition probabilities from each state must equal one. For instructions regarding how to add, remove, and rearrange transitions, and make use of additional functionality available on this page, and others like it, click here. Specifically, in building a model with many possible transitions, it may be useful to make use of the row grouping functionality—for example, 'Group by From' will organize all the transitions you have created into groups based on each 'From' state. This functionality allows you to limit your view to transitions of interest—to ensure that you have defined all of the necessary transitions for each 'From' state and that they sum to one.
3-State Partitioned Survival Model
In a 3-State partitioned survival model, the transitions page defines the distributions of PFS and OS for each strategy. This is done using a table in which each row represents the PFS or OS for a given strategy. This table has the following columns:
|
Column |
Description |
|
Strategy |
The strategy for which the distribution is for. |
|
Endpoint |
Whether the given distribution represents PFS or OS. |
|
Unit |
The time unit in which the given distribution is defined. Can be either 'Days', 'Weeks', 'Months', or 'Years'. |
|
Formula |
A formula which defines the survival distribution used. |
For a 3-state partitioned survival model, there is no need to add, remove, or reorder rows in the transitions table. Each strategy automatically will have one row for PFS and one row for OS.
Custom Partitioned Survival Model
In a custom partitioned survival model, the transitions page defines the calculations of the probability of being in each state over time (i.e. the trace probabilities).
|
Column |
Description |
|
Strategy |
The strategy for which the trace probabilities are being calculated. In order to define a trace calculation which is shared by all strategies, select 'All'. |
|
State |
The health state for which the trace probabilities are being calculated. |
|
Formula |
A formula which defines the calculation of the trace probabilities. This formula should generally be a function of model time. The C keyword can be used to automatically calculate the complementary probability (i.e. one minus the probability of being in all other states). |
Please note that for the transitions page to be valid for a Custom partitioned survival model, there must be one row defined for each health state for each strategy and the probabilities of being in each state must sum to one for each strategy in each cycle. For instructions regarding how to add, remove, and rearrange trace probability rows, and make use of additional functionality available on this page, and others like it, click here.
