Making the Most of Drug Development Data

The survey indicates that drug development organizations are still a long way from establishing data context. In fact, most companies continue to take a flat, document-centric approach, rather than the multi-dimensional views essential for successful drug development today.

Some organizations are taking marginal steps to improve their development team members’ ability to use documents, for example, by offering keyword search capabilities. But fewer than 5% of respondent say that their organizations are taking advantage of structured databases (Figure 4), while 15% indicate that they have no structured data capabilities at all.



Nobody could clearly identify structured data benefits offered by their existing IT system (Figure 5, below). When asked about “drill down” and other capabilities, only 20% of the 90 people who responded to this question could name two or more characteristics, and only 5% could answer that they had four or more of the characteristics listed.

These characteristics maintain the context under which information is generated, and correlate data across key development parameters (who, what, where, when, how) for the entire development lifecycle, allowing practical planning, execution and analysis to take place. While “drill down” capabilities take the user to the right frame of reference for the detailed search — whether by time, date, project or phase — “drill across” capabilities allow users to develop the relationships between processes, materials, equipment, environment, personnel and operating modes (e.g., GMP vs. non-GMP).



Enabling this context and maintaining the correlation allows development users to follow the thread of information as it is revised, ensuring that all changes are made for the right reasons and determining the impact that such changes could have on overall product performance. Maintaining correlation also ensures that the history and evolution of changes made in a process route or a material can be captured and accessible throughout the development life cycle. Together, these capabilities allow users to pull up and understand data that may have been generated several years ago, ensuring that yesterday’s lessons are applied to later-stage commercialization studies such as comparability studies and campaign summaries.

Responses (Figure 5, above) show that many companies don’t have the right infrastructures in place to realize the full benefits of structured data. This lack may prevent them from achieving the goals outlined by FDA’s Critical Path and PAT initiatives, both of which aim to foster a culture of doing things “right the first time” and being more predictive by leveraging lessons from the past.

Only a few respondents (between 7 and 20 out of 104) have systems in place to help users avoid mistakes, proactively — for example, notifications that can help an operator make necessary adjustments before the issue becomes an irreversible problem or becomes a reportable error.

Size doesn’t matter!

Interestingly, both ad hoc and data silo practices occurred in companies regardless of size. Relying on unstructured data and having limited proactive decision-making capabilities appears not only to prolong the time required to search for data, but leads to unnecessary duplication of efforts — for example, repetition of experiments. Figures 6 and 7, below, indicate the average time that individual researchers or development professionals spend looking for data, along with the percentage of time that data cannot be located at all. On average, respondents spend almost five hours per week (or nearly 13% of their time) looking for data for reports that they must prepare, whether progress reviews, filings, technical summaries or engineering studies.



Two-thirds of the participants could not find the information they needed approximately 10% of the time. A combined 28% of respondents indicated that data could not be found between 20% and 60% of the time.




This inability to find important data puts any organization at enormous regulatory risk and drains the scientific and engineering lifeblood of drug development. But it also reduces pilot plant capacity utilization, and increases the cost of running lots and batches due to the number of experiment “reworks” or repeats that have to be executed to fill various information gaps that may exist.

Figure 8, below, shows how it leads to unnecessary “rework”, even during late-stage process characterization and commercialization. 63% of the participants said they had to repeat experiments at least 10% of the time and 31% reported having to duplicate efforts 25% of the time. Overall, 8% of the experiments or tests must be repeated when data cannot be located, a significant direct cost, but an even greater opportunity cost, particularly for pilot plants that are running near capacity.