Cenduit_Whitepaper_Forecasting_Present_14June2016

As the world’s leading IRT specialist,
We see things others don’t.
Clinical Supply Chain Forecasting
and Simulation
Cenduit White Paper Series: Issue 4
Accurate forecasting of clinical supplies has become increasingly important in the past several years as planning the
amount of study drug and other materials required to complete a trial has grown in complexity. Key drivers adding
to this complexity include increased numbers of trials, reduced start-up timelines, patient recruitment challenges,
globalization leading to more distant clinical sites and a variety of regulatory requirements, and escalating costs
for investigational drugs and comparators.[1]
To meet these challenges, sponsors
have turned to a variety of forecasting methods ranging from manual spreadsheet
calculations to software that enables sophisticated supply chain modeling.
Well-designed forecasting solutions provide real Business Intelligence (BI) that
enables data-driven decisions when planning, running and closing a study. And
IRT Systems provide the ideal platform to collect and feed real-world information
into the forecasting solutions.
When planning a trial, forecasting solutions help sponsors to determine the
amount of medications and other supplies they will need to manufacture and
package. Using modeling techniques, sponsors and CROs can also plan an effective
supply chain strategy. Forecasting is also valuable during the execution of a study
Planning the amount of study
drug and other materials
required to complete a trial
has grown in complexity.
Key drivers include increased
numbers of trials, reduced
start-up timelines, patient
recruitment challenges, and
globalization

to make sure that the supply chain strategy remains on track and when necessary, allows sponsors to take
proactive measures to ensure ongoing success. As a study is drawing to a close, forecasting helps to ensure not
only that there will be enough supplies to complete the trial, but also that a minimal amount of medication will
be left over at sites and depots, thereby minimizing wasted supplies and the associated costs. At times sponsors
and vendors have also employed simulation techniques for the same purposes.
This paper will discuss how clinical supply forecasting is typically used in planning and running a trial. It will also
describe what is typically meant by the terms “simulation” and “forecasting.” First, it is important to establish the
main goals of simulation and forecasting.
Goals of Clinical Supplies Forecasting and Simulation
The most important objectives of forecasting and simulation are 1) to plan for the initial supplies that will be needed
for a trial, 2) to optimize the entire supply chain strategy, 3) to ensure that the supply strategy remains on track
until the trial is complete and 4) to minimize and mitigate any risk in the supply chain (see Table 1). Forecasting
Cenduit Forecasting White Paper
2
Table 1: Goals of Forecasting and Simulation

3
enables Clinical Supply Managers (CSMs) to make data-driven decisions and to gain insights into the supply needs
of a trial by exploring various scenarios. It also allows them to change various parameters to see what may happen
in the real world. For example, a CSM may run a forecast to see if adding new sites to the study will cause a shortfall
in any supplies. The CSM may also model a number of “what-if” scenarios to see what might happen if sites in
some regions recruit patients rapidly while other regions have slower enrolment. By modeling the future this way,
supply managers gain insights into the supply chain and can proactively make adjustments before issues arise.
Defining Forecasting and Simulation
Forecasting refers to techniques to predict future outcomes based on past events and professional insights gained
from observation. Methods usually rely on mathematical formulas based on facts, observed trends, data models
and at times, expert opinions. Perhaps the simplest analogy is the weather forecast. There are different weather
models which generally produce similar, but not identical, predictions. This is evident when one checks several
different websites for the next day’s weather and minor variations are observed. Of course, the farther into the
future one looks, the more the predictions of each model will vary and the more the reliability is reduced.[2]
So when planning an important event, it is wise to check the forecast regularly as the day draws nearer. The same
holds true in trial supply forecasts.
As a clinical trial is launched into the real world, many of the assumptions used during the planning phase, including
any pre-study simulations, may change. For example, if recruitment rates are slower than anticipated, the study
will experience delays and will likely run longer than anticipated. Will some drug expire during the study as a
result? Will another packaging run be required? Will more sites be needed to
recruit additional patients? These are just a few factors that will affect the supply
chain and the initial forecasts.
Simulation is a means to generate forecasts and to reveal meaningful details about
the system being modeled. It is used to model the real world using probability
distributions and allowing for uncertainty by using ranges of values for factors that
may vary. Then the model is run many times with each iteration producing different
results due to the built-in variability of some parameters. The results of each run
are then saved and aggregated to produce a report that typically shows the average
(mean and/or median) outcome, standard deviation, range (highest and lowest
values) and similar descriptive statistics. Such reports provide insights into the
system being modeled. As a side point, the weather forecasts mentioned earlier,
are actually the results of simulations; however, only the most probable forecast from the simulations is reported in
the form of the expected high and low temperature, humidity level and pressure. Details, such as the temperature
ranges and potential variations observed in the simulation results, are omitted.
To further illustrate how simulation works, suppose we were setting up a model to simulate coin tosses. We would
set up the simulation with an equal probability of getting a head or tail on each toss. If we run the simulation once
Simulation is a means to
generate forecasts and to
reveal meaningful details
about the system being
modeled. It is used to
model the real world using
probability distributions
and allowing for uncertainty
by using ranges of values for
factors that may vary

4
and the result is “heads,” that would represent one possible outcome, but would provide little additional
information. Now suppose we simulate 100 coin tosses and aggregate the results. We could then report the
average number of iterations where the result was heads along with the standard deviation. We could also answer
other questions such as: What was the highest number of consecutive heads? How often did the simulation
generate two consecutive tails? Thus, even this simple simulation experiment would provide information about
potential outcomes that may help to anticipate what would happen in the real world.
Applying these concepts to the clinical supply chain, a simulation model may be set up with a number of param-
eters, some allowing for variability and others fixed or predetermined. Figure 1 shows the main categories and
key factors that would be included in a forecasting or simulation model. Some fixed parameters would include the
numbers of depots, sites and countries and the study visit and dosing schedule. Other parameters will allow for
variation between runs. Variable factors would include patient recruitment rates, shipping times between
locations and site activation rates. For example, the model may involve varying patient recruitment rates for high
and medium enrolling sites. A high enrolling site might expect between zero and two patients per week during the
recruiting period, while a site with a medium rate might expect between zero and one. During each simulation run,
Analytics & Reporting:
Status & Forecasting
• Strategic Dashboard with live data
• Intelligent Data Mining
• Real-time Material Forecasting
Site and Country Parameters
• Number of Sites
• Recruitment Rates
• Lead Time into Countries
Study Design
• Randomization Algorithm
• IRT Configuration Parameters
• Adaptive Designs/Titrations
Depot Planning
• Depot Hierarchy
• Depot Supply Parameters
• Lead times (depot to depot,
depot to site)
Packaging Configuration
• Expiry Date Management
• Subject Dosing Schedules
• Pack Types
Figure 1: Key Factors in Trial Supply Management

5
the model would then randomly generate and assign patients to sites based on those rates. Next the model would
be run through many iterations and the results would be aggregated allowing the sponsor to answer questions
such as: How long did it take, on average, to recruit all patients into the trial? What were the fastest and slowest
recruiting times in the results? Were there any stock-outs at depots or sites in any
of the runs? What was the average amount of drug left at each depot and site?
The answers to these and similar questions can be valuable in setting up the
clinical supply chain and in making sure that it remains on track during the entire
trial. Since the supply chain for most modern studies is automated using an
Interactive Response Technology (IRT) platform, forecasting results are also very
useful in setting up and configuring the IRT parameters and the real-world data
collected by the IRT can shape ongoing forecasts or new simulations.
Forecasting with Interactive Response Technology (IRT)
IRT systems have become indispensable tools for managing the clinical supply chain in an efficient and cost-
effective manner. These systems automate the complex processes involved in Randomization and Trial Supply
Management (RTSM) and enable sponsors to save money by minimizing the amount of overage required to run
a trial while simultaneously reducing the number of costly shipments required to move supplies from depot to
site. Achieving such efficiencies requires balancing the very same supply chain factors discussed above and shown
in Figure 1.[3,4]
When running a pre-study forecast, it is good practice to make sure that these factors are aligned
with the IRT system configuration to get the most accurate results possible; however, at times the sponsor may
be planning a manufacturing run even before they have selected their IRT vendor. This is an important reason to
perform ongoing, mid-study forecasts. Additional insights may be gained by repeating the pre-study simulations
using real-world data to update the assumptions made before the study began.
There are additional reasons to perform mid-study forecasts. For example, while some sponsors have developed
their own supply forecasting methods, many are simply spreadsheet calculations that are used to determine the
overall amounts of study drug and comparator that will be needed. Such basic approaches have value when
planning, but they usually have the several limitations. First, they are detached from the actual supply chain
methodology that will be built into a well-constructed IRT system and therefore, do not consider how the system
will optimize the management of study supplies. Second, they rely on baseline assumptions about the study that
may change, such as the expected recruitment rates, number of sites, study duration and so on. Third, many of
these spreadsheet approaches ignore vital factors like expiry date management. For these reasons, more rigor is
needed when forecasting supply needs.
Modern IRT platforms generally include features like built-in intelligent forecasting, standard and ad-hoc reporting
capabilities as well as dashboards and analytics that allow CSMs to explore future supply needs by modeling
“what-if” scenarios. As a study progresses, the IRT holds critical, real-time information about the entire clinical
supply chain. The system not only tracks where every unit of study medication and other supplies are, but is
Forecasting results are also
very useful in setting up and
configuring the IRT parameters
and the real-world data
collected by the IRT can shape
ongoing forecasts or new
simulations.

6
also aware of the future supply requirements based on the visit and dosing schedules, how far each patient has
progressed in the trial, whether any supplies have been damaged in transit and so on. When this information is
exposed to a supply manager via an interactive forecasting dashboard, it creates a powerful and valuable tool
enabling teams to take a proactive approach to managing supplies.
Figure 2 provides an overview of the value of forecasting in the planning and execution of a trial; it also shows
some of the key areas that are typically analyzed and, if necessary, updated. For example, a CSM viewing the
enrollment reports may notice that recruitment is lower than expected in the US, but ahead of schedule in Eastern
Europe. If this trend continues, will the drug supply sent to the Eastern European depot be sufficient? A quick
Figure 2: Value of Forecasting in Planning and Execution of a Clinical Trial

7
Figure 3: What is a Forecasting Dashboard?
Simply put, a Forecasting Dashboard is an interactive, single-page view of the clinical supply chain as it looks at this
instant in time and based on current trends and user controls, how it is expected to look in the future. The dashboard
conveys this information at a glance with easy-to-read graphs, charts, tables and visualizations. It also allows provides
straightforward controls that allow the user to explore what will happen if specific actions are taken or if certain trends
change. For example, the user may model the future by dragging a slider to add investigative sites or to increase or
decrease patient recruitment rates.
Other vital features of a solid Forecasting Dashboard include the ability to look at the supply chain across multiple
trials (pooling supplies) and to drill down to specific regions, countries, depots, sites and so on. The user should also
be able to filter for specific types of materials or medications with a simple click or screen tap. Finally, the user should
be able to save and share the scenarios and views created to help support and document supply chain decisions.
In the sample screenshot shown in this figure, the user has checked the “Forecast” button and is exploring several
what-if scenarios (rather than simply viewing the current status). Using the slide-bar controls, the user is modeling
what would happen if two more sites were activated, if another site per month is added and if the targets for total
patients randomized and number randomized per month are increased. As shown on the line graph, these proposed
changes would immediately affect the current stock levels as additional medication kits would have to be sent to the
newly activated sites and more kits would be needed at existing sites to handle the increased randomizations. (For
example, note that the stock level for medication type “5mg 3 day” falls from current level of 520 kits to about 420
as shown on the middle blue line of the line graph.) Further, this scenario would lead to a stock-out of at least one
medication type by September.

look at the forecasting dashboard will provide the answer by showing how long the depot supply will last based
on the current recruitment rate. More than that, the CSM can drill down to individual sites in the region to make
sure the current supply chain strategy will avoid stock-outs at the site level. If a stock-out is predicted, then the
user can set up additional scenarios to see what will prevent the shortfall. For example, they may set up models to
answer questions such as: “If I move supplies from the USA depot to the Eastern European one, will it prevent the
stock-out without creating a similar issue at USA sites? If the recruitment rate slows just a little, will we still have a
stock-out? Will we need another manufacturing run?” (For details, see Figure 3: What is a Forecasting Dashboard?)
By regularly checking the forecasting dashboard, the CSM and other study team members can proactively handle
any potential issues to keep the study supplies on track. They can also make sure that the IRT will continue to
supply every site and patient exactly as needed without wasting supplies.
Conclusion
Planning, tracking and managing the clinical supply chain is critical to the success of any clinical trial. Further,
there are dozens of factors and decisions involved in the ongoing health of the supply chain. Forecasting and
simulation are valuable tools that help a CSM to perform critical planning activities,
such as optimizing the IRT supply chain parameters and making sure ample
supplies are on hand to run the trial. In turn, a fine-tuned supply chain strategy
reduces costs by making efficient use of supplies, reducing waste that is often
caused by overages and by allowing just-in-time shipments when medication is
very expensive or in short supply. Once the study goes live, the CSM can gain
important insights into the supply chain by using current, real-time information
directly from the IRT to look at updated forecasts. If issues are predicted, then
different mitigation plans can be modelled proactively in the forecasting dashboard
to guide the CSM to an effective solution. In this manner, forecasting dashboards
provide meaningful support for data-driven decisions that help to ensure the health, costs and overall performance
of the entire clinical supply chain.
8
Cenduit Forecasting White PaperCenduit Forecasting White Paper
Planning, tracking and
managing the clinical supply
chain is critical to the success
of any clinical trial.
Further, there are dozens of
factors and decisions involved
in the ongoing health of the
supply chain

As the world’s leading IRT specialist, we see things others don’t.
•• Bringing together expertise, technology, innovation, and vision
•• Meeting all randomization, trial supply, and patient management needs
•• Seeing exactly what our customers need and how to help them succeed
Contact us, and learn how Cenduit sees things others don’t. email: corp.communication@cenduit.com web: www.cenduit.com
Endnotes
[1] Lamberti, Mary Jo, PhD, “Global Supply Chain Management", Applied Clinical Trials, September 1, 2012.
[2] Kwon, Diana, “Are Europeans Better Than Americans at Forecasting Storms?”, Scientific American, October 1, 2015.
[3,4] “White Paper: Factoring the “What Ifs” into Supply Forecasting: Why Building a Durable Supply Chain Around a
Protocol Is Critical”, Fisher Clinical Services, fisherclinicalservices.com.

Cenduit_Whitepaper_Forecasting_Present_14June2016

More Related Content

What's hot (19)

Similar to Cenduit_Whitepaper_Forecasting_Present_14June2016 (20)

Cenduit_Whitepaper_Forecasting_Present_14June2016