Knowledge Base for Supply App
IRT Configuration (1)
Table of contents
IRT configuration
In the IRT configuration tab, the user specifies all the IRT settings that will be used for the simulations. These settings are choices made when setting up the trial with the IRT provider, and will impact how the IRT resupply algorithm will work.
If the trial is starting in a few months, the user can try out multiple scenarios playing with these settings to help design the IRT system in the way that best suits the trial.
If the trial is already ongoing, the user should ask for information from the IRT team in order to choose the right settings that will allow the Supply App to replicate the actual IRT behavior.
IRT resupply algorithm
The user can choose between a min/max (buffers only) or a combined (buffers and prediction) strategy. This choice applies to sites only, as depots resupplies are not managed by the IRT.
The min/max strategy will only use buffers to cover the patient demand on site. Nothing will be predictable, so the buffer stock will be quite high.
The rule for resupplying the sites will be really simple (taking as an assumption that shipment triggering is based on stocks only, see below): if the inventory considered at site is lower than the min buffer level (or trigger level), a shipment will be triggered. The amount of kits sent to the site will be the max buffer level (or resupply level) minus what is already considered on site.
# kits on site + # kits in transit towards the site – # kits past the do not count date.
The combined strategy will use both buffers and prediction, meaning that part of the patient demand will be forecasted inside the IRT system, instead of relying on buffers only. Buffers may still be necessary to cover randomization, titrations and diverse unscheduled events. Prediction will be used to cover demands such as patient dispensing during maintenance phases, when the visit interval is longer than the lead time to site.
The rule for resupplying the site will be a little more complex than for pure min/max strategies. There are two additional concepts to take into account:
- The short window is the number of days for which the inventory should cover the worst case for the demand.
- The long window is the number of days for which kits should cover the predicted demand when a shipment is triggered.
min buffer/trigger level + predictable demand during the short window.
When the shipment is triggered, the amount of kits sent will be:
max buffer/resupply level + predictable demand during the long window – inventory.
Single do-not-ship per shipment
This parameter will impact how kits are managed around expiry replacement, in the case of trials where the visit intervals are not the same throughout the treatment. In those cases, there is a distinction to be made between buffers and prediction: buffers should be valid for any dispensing that may come, so the validity required for buffer needs will be driven by the longest visit interval possible. Prediction on the other hand is visit-specific. Indeed, the IRT system knows for which visit each kit is predicted, and it is then possible to require a validity that is driven by the specific visit for which the kits are predicted.
The question is: will the IRT use that possibility, or also require a validity that covers the longest interval?
If not, kits that are ordered for prediction needs will have a required validity that is dependent on the visit interval that they are predicted for.
Use short window for computing safety stocks
This option will be used for the computation of buffers if the user does not define the buffer levels.
If the option is not selected, the min buffer (or trigger level) will be computed to cover the worst case demand over the lead time to sites.
As the short window has to be greater or equal to the lead time, the first option is a little more conservative.
Group shipments at the end of a trial
At the end of trials, it is common to have very few patients remaining, implying a very small demand in terms of kits to be dispensed or shipped to sites and depots. This option allows the user to group the last few shipments to be sent to depots, in order to have one shipment only in the last few months instead of multiple ones with small quantities.
Indeed, these shipments will cover the demand worst case but this worst case depends on the chosen safety stock confidence level (SSCL). For example, if a SSCL of 98% is chosen, 2% of the worst cases will not be covered because considered too extreme / unrealistic by the user. It is advised, most of the time, to use a SSCL of 99% or 98% in order not to cover these extreme cases while still being conservative enough.
If risk appears, it is recommended to turn off this option.
Shipment triggering
The user chooses if shipments are triggered by inventory levels only (stock trigger), a time criterion (time trigger), or both at the same time (stock time trigger).
The Stock trigger criterion will use the rules described above, in the IRT resupply algorithm parameter. Comparison between the inventory level at site and the required level (including buffers and prediction, if used) will be the only rule defining if a shipment has to be triggered or not.
The Time trigger criterion will use a shipment triggering rule based on frequency. A shipment will be triggered if a full prediction window has passed since the last shipment was raised. The prediction window is the interval between the short window and the long window. It is defined in the Resupply management tab of the IRT setup.
The Stock time trigger criterion will use a mix of both algorithms. If the stock trigger criterion or the time trigger criterion is met, a shipment will be raised.
Trigger by time rules is quite uncommon, so it is recommended to use the Stock trigger option.
