Quality by design across clinical development
Quality by design builds quality into everything a sponsor does in clinical development from the outset, while the industry has relied on testing quality in afterwards. It rests on knowledge of the process and on measured quality, and moving to it calls for a quality system that is integrated, risk-based and overseen at board level.
What quality by design is
Quality cannot be inspected into a trial; it must be built in from the outset, in the trial's protocol, design and planned operational conduct. Quality by design is standard practice in manufacturing. The FDA wants trials to use it to avoid excessive regulatory oversight, which is expensive and comes too late: by the time of an inspection the trial may have been over for a long time, so the regulators can find out what went wrong but cannot guarantee quality. The regulators would rather avoid that oversight by having quality built in from the beginning, through a systematic and proportionate approach to clinical development that emphasises process control, and that approach is what quality by design is: process control. Monitoring and auditing remain important quality tools, but they must be flexible and should not be the only approach.
Quality by design is one of the foundational elements of risk-based quality management, and it focuses on making sure the trial is built with quality in mind. It forms part of ICH E8(R1), and in 2023 it was coming in ICH E6(R3) as well, so discussions of quality by design were expected to become more prevalent.
Quality by design relies on profound knowledge, Deming's term for the experience people hold, and knowledge of the process is important to build in. It is similar to the plan-do-study-act cycle, with every study run iteratively through the cycle. It focuses on risk-based approaches, not only in monitoring but throughout everything done in clinical development. It encourages continuous quality improvement, which needs quality to be measured and defined. Quality by design is about measured quality.
Its intention can be put as designing reproducible processes, where Six Sigma speaks of designing capable processes. Once quality is defined, it can be known whether trials are reproducible over time. That matters because the FDA expects two identical phase 3 studies to show that results are reproducible. In 2021 achieving quality by design was still some time off, but with a really strong implementation of it one study might be enough.
Where the industry's quality practice stood
The European Medicines Agency regarded the industry as testing quality in. One practitioner who reviewed almost 1,600 standard operating procedures from different companies over five years found procedures and systems built up by patching things together rather than by robust implementation. In quality terms the industry stood where a 1950s car factory stands, relying on auditing, inspection, monitoring and retrospective document checks, and on perfection in its plans.
Organisations were far too siloed, so their data were not integrated. Quite often data managers had no access to the on-site monitoring system, and on-site monitors none to the data management system. Standard operating procedures were not integrated either. Companies had too many plans, some more than 25 for one study, so the plans were put in the cupboard and not read, and non-compliances followed. Risk management was used only to a limited extent, and oversight at board level was lacking, with boards all too often looking only at budgets and timelines.
Asked in a 2021 poll of medical-writing webinar attendees whether their company had developed a quality by design concept, 56% of respondents did not know, 26% said no and 19% said yes. Many companies that said they were doing quality by design did not in fact have a full concept of what is needed, and used it in parts of the organisation rather than across the whole of it. Until recently many companies were not using quality tolerance limits, so when quality by design came along there was a real danger that they would develop something else rather than use the existing tools the regulators had given them.
The quality system to move to
At a European Medicines Agency meeting on ICH E8 and ICH E6(R3), the head of a European cancer research organisation said to forget quality as it is known. One experienced practitioner reads this as saying that current quality methods are not the ones to use and the way things have been done has to change.
The industry needs to reach an automated process, though its work can never be totally automated. The quality system it needs has these parts:
- Quality that rests on defined and measured quality, not on perfection.
- Quality tolerance limits used not just as a tick box but throughout the organisation, with an understanding of why they are used.
- Continuous quality improvement once quality is measured, and then predictive quality based on it.
- Tools to avoid being a siloed organisation, with data systems integrated from source to submission, which is where the regulators want the industry to be: everything together rather than having to do computer system validation on data transfers.
- Risk management as the basis of all the work.
- A defined quality philosophy that everyone follows. One experienced practitioner prefers the plan-do-study-act cycle for it, because in their view everyone can implement it.
- Comprehensive oversight, with the board using safety and data integrity metrics as predictors of timelines.
One experienced practitioner holds that if the only metrics a board receives are budgets and timelines, an inspector will conclude that the company does not value quality or safety, because it is not looking at them at a high enough level, so a company has to be aware of the perception it gives to others.