Modeling IVR Studies Ship Times
Posted on Wed, Aug 18, 2010 @ 09:56 AM
Modeling IVR Managed Studies that Ship Only on Certain Days of the Week
Jeremy Daniel, Senior Consultant
We often run into cases where an IVR (interactive voice response) system may not generate shipment orders for study kits every day. For example, commonly it seems they may only ship Monday through Thursday. In many cases this could be due to cold chain supplies having to be received immediately and being unable to guarantee that someone will be there to receive the shipment on a weekend. Others, however, may be for unrelated policy reasons. Standard modeling practice says to add a day to the shipping lead-time for every day that is not available for shipping (3 days in the case of the above example). Related to this concept, there is often some confusion for new users about whether the shipping lead-time for the IVR study refers to calendar days or business days.
One of the nice things about an IVR simulator (for instance, BioClinica Optimizer) is the ability to see the effects of changes in the resupply algorithm on your study. So, what is the impact on the results of the simulation caused by this simplification of adding days to the shipping time to represent days shipments are not made? To test this, I took a simple IVR managed study with 2 treatment groups and 600 patients in 14 centers. Sites are supplied with kits for 2-3 patients and each subject has 5 predictable treatments after randomization. On the baseline scenario, shipments take 6 days to get from depot to site; the comparison uses 3 days with no shipments on Friday through Sunday*.
Using this sample IVR managed study model, I ran a number of monte carlo simulations and found only minor differences. There were about 5% less mean shipments made by the comparison scenario (417.61 compared to 438.77). As would be expected, since the original scenario was not fully optimized, there was also a lower mean lost subjects due to stock out (0.22 compared to 0.99).
What have you found in your IVR studies to be the impact of not shipping every day?
* In both versions, we have a maximum of 6 days from depot to site so I used a 6 day short window (triggers shipment if supply will be short in the next 6 days). The comparison uses a script to stop shipments by setting the floor (minimum stock at site) to 0 on days that shipments are not made and setting the short window demand (patient need for 6 days) to 0 as well.