By Pirmin Froehlicher

October 4, 2016 | Have you ever seen the meme of the expected path (direct line from A to B) and then the real path of convoluted squiggly lines and diversions? The latter is clinical drug supply. In an ideal world, there would be just enough drug at each clinical site around the globe to ensure the study runs smoothly. No wasted supply and no risk of trial disruption.

Unfortunately, this is not an ideal world but rather the real world. Clinical trials are becoming more complex as new designs are introduced. Patient enrollment can vary widely within regions, countries and sites due to competitive enrollment and other variable factors such as discontinuation rate, titration probabilities, weight or Body Surface Area (BSA) of patients which directly impacts dispensing. Since high enrolling sites need more product than low enrolling sites, the typical response from a clinical drug supply manager would be to define a handful of enrolling categories to which sites are assigned, e.g. low, med, high, and very high. These categories typically define the minimal threshold and the initial quantity sent, which can cause waste in sites that do not enroll patients. The challenge in clinical supply forecasting is to set those values as low as possible but as high as necessary, taking into account all variable factors as well as required stocking of the depots, expiration date and the overall availability of the clinical drug.

So what options are available to clinical supply professionals to address that challenge? Sure, supply chain forecasting tools are available to help support more complex trials. However, these tools can be complex in and of themselves. Unfortunately, clinical drug supply managers that only have to forecast a few studies never really get the chance to gain experience in using them, optimally. In essence, clinical supply forecasting is challenging to get right. Given that, is an approximation enough?

An experienced clinical supply manager can come up with good estimations, with buffer quantities, and manage the supply with a certain level of confidence. A majority of the industry relies on their supply manager’s expertise – rightfully so – and uses spreadsheet planning or more sophisticated average planning that a typical ERP system can support. After inputting the expected number of patients in each country, the milestones, dispensing schedule and treatment arm ratio, plus potentially additional information depending on what the system supports, the system will calculate an average demand which can be a stable method of planning on study level. Planning in ERP is also attractive because it is integrated with other parts of the clinical supply organization such as distribution, order preparation and execution, materials management and even commercial ERP systems. It does not, however, address the corrupting influence of variability, i.e., in what regions which patients are going to appear, what treatment arm they will go into, or the titration, etc. Average planning will not account for those variables and it does not provide any support in defining optimal buffer levels.

To address those gaps, clinical supply simulation tools can be used. Those tools account for the variability in clinical studies by executing hundreds or even thousands of simulation runs and the output also allows to actually optimize clinical supply and support risk-based decisions. How much stock do I require with a 99% confidence level, how much with 98%? Is the saved medication worth the risk? Or does a lower service level even allow me to start a study with supply constraints that would otherwise need to be delayed? A way of overcoming the complexity of using those tools in the right way is to have a dedicated user group. In a larger organization, the savings of this should easily outweigh the costs and in smaller organizations, a few supply mangers may specialize in optimization.

Two large challenges remain. One, receiving accurate input information for planning and optimization. This is key not only before study start but also during trial conduct. And two, making sure that the planning and optimization that took place in whatever tool is based on the actual execution processes. This is especially true for the processes and algorithms used by RTSM which steer the inventory and re-supply situation at the sites, and potentially depots.

I’ve seen supply managers come up with drug supply planning on a piece of paper and then double it to offset risk. Is it wrong? Not necessarily. Does it almost ensure there will be supply waste? Absolutely. So, how can supply planning professionals stop choosing between having clinical supply waste vs. supply chain risk to inform forecasting decisions? In summary, I am proposing five best practices for clinical supply professionals to tackle this issue.

Best Practices for Clinical Supply Forecasting

• Garbage in equals garbage out. Supply forecasting can only be as good as the input data received. This is important to understand for both supply professionals and clinical operations. Ensure data is clean and accurate before filtering through a supply forecasting tool.
• Continually update input data as the study evolves to adjust the supply strategy—or better yet, integrate actual enrollment data into the supply planning tool
• Use sophisticated simulation or statistical calculation tools to address the corrupting influence variability in clinical trials
• Consider having a small dedicated user group for those tools to build a solid knowledge base and use them in an optimal way
• Close the gap between the way you forecast and the way your RTSM works. The closer you are to the way your system works, the better.

Leverage these tips to help your organization straighten the line in clinical drug supply, ultimately reducing waste and mitigating supply risk.

Pirmin Froehlicher is Client Services Lead at 4G Clinical. He has 6 years of experience in clinical supply and extensive knowledge in clinical supply forecasting and optimization, and sound knowledge of RTSM technologies. Pirmin holds a BS in Industrial Engineering and a certification as Lean Six Sigma Green Belt. He can be reached at pirmin@4gclinical.com.