Process analytical technology is the part of pharmaceutical quality that most engineers can point to: a probe in a reactor, a spectrum on a screen, a chemometric model that returns a number every few seconds. Quality by Design is the framework that explains why the probe is there at all. It is the older idea, the broader one, and the regulatory parent under which PAT sits.
QbD is also the part of the regulatory landscape that newcomers to process analytics most often skip. The temptation is to read the 2004 FDA PAT guidance and treat it as self-contained. It is not. PAT was written as one mechanism inside a larger reform programme, and the framework that programme built - articulated through ICH Q8, Q9, Q10, and Q11 - is what gives PAT its regulatory weight.
This article is a primer on QbD for readers who already know roughly what PAT does and want to understand the framework it serves.
Where QbD came from
The phrase “quality by design” predates pharmaceutical regulation. It was used by Joseph Juran in the 1980s to describe a quality philosophy in which finished-product testing is the last line of defence, not the primary one. The idea reached pharmaceutical regulation through the FDA’s 2002-2004 reform initiative, Pharmaceutical CGMPs for the 21st Century - A Risk-Based Approach. That initiative had two visible outputs: the 2004 PAT guidance, and a final report that set out a long-term goal of moving the regulatory model from prescriptive recipe-following to science-based process understanding.
The International Council for Harmonisation absorbed the same goal into its quality guideline series. ICH Q8 (2005, revised 2009 as Q8(R2)) translated QbD into a written framework. ICH Q9 (2005, revised 2023 as Q9(R1)) covered the risk-management mechanics QbD relies on. ICH Q10 (2008) set out the pharmaceutical quality system that holds everything together. ICH Q11 (2012) applied the same logic to drug substance manufacture. The four documents are the core QbD reference set; later guidelines (Q12 on lifecycle management, Q13 on continuous manufacturing, Q14 on analytical procedure development) extend rather than replace them.
The QbD vocabulary
A reader new to the framework should learn five terms.
The Quality Target Product Profile (QTPP) is the description of what the finished product needs to deliver clinically and physically: route of administration, dosage form, strength, stability, key performance attributes. QbD begins by writing it down explicitly.
A Critical Quality Attribute (CQA) is a physical, chemical, biological, or microbiological property of the product that needs to be within a defined limit to meet the QTPP. Particle size for a dry-powder inhaler. Dissolution profile for an immediate-release tablet. Aggregate content for a monoclonal antibody. CQAs are the things you actually measure or control.
A Critical Process Parameter (CPP) is an input or process variable whose movement affects a CQA. Granulator impeller speed. Crystalliser cooling rate. Bioreactor dissolved-oxygen setpoint. Identifying CPPs is the work of process understanding - the part of QbD that justifies why you decided what to measure.
The design space is the multidimensional region of CPP and material-attribute combinations that has been demonstrated to produce a product that meets the CQAs. Movement within the design space is, in regulatory terms, not a change; movement outside it is. The concept is one of QbD’s central practical contributions: it lets a manufacturer adjust a process within agreed bounds without filing a regulatory change.
The control strategy is the documented set of controls - input specifications, in-process controls, finished-product tests, monitoring systems - that together ensure the process operates inside the design space. PAT instruments, when present, are part of the control strategy. So are conventional offline tests, batch records, and personnel-training programmes.
How PAT fits inside QbD
In QbD terms, a PAT instrument is one option among several for executing the control strategy. The framework does not require PAT, and it does not privilege any particular measurement technology. What it asks is that the manufacturer can demonstrate, with data, that the chosen controls are adequate for the identified CQAs and CPPs.
In practice, PAT becomes attractive whenever the CPPs move faster than offline testing can characterise them. A continuous granulation line cannot wait for an HPLC result; a fermentation cannot be paused for a four-hour assay; a crystallisation endpoint cannot be detected by a balance. Inline NIR, Raman, FBRM, and FTIR each addresses a class of these problems and produces data on a cadence the process can use. The choice of which to deploy is, in QbD vocabulary, a control-strategy decision. Our Raman vs NIR decision framework walks through one of the more common comparisons.
Inside the regulatory file, the PAT instrument’s place is documented through the same machinery as any other in-process test: a procedure, a validated method, and a defined link to a CQA. The 2011 FDA process-validation guidance reframed the lifecycle in QbD-compatible language - Stage 1 process design, Stage 2 qualification, Stage 3 continued process verification - and that lifecycle is where PAT lives in operations.
QbD beyond pharma
QbD’s vocabulary is pharmaceutical, but the underlying logic moved into adjacent chemical industries well before the regulatory framework existed. Polymer producers, specialty-chemical houses, and food and beverage manufacturers have used the same loop - identify the attributes that matter, identify the parameters that move them, deploy measurements that report on both - for decades. The chemical industry tends not to use the words CQA and design space, but the practice is recognisable.
What pharmaceutical QbD adds is the regulatory mechanism. A polymer producer is free to redefine its control strategy as it learns; a pharmaceutical manufacturer must file a change. The design-space construct is, in part, an attempt to give pharmaceutical operators a defined zone of operational freedom equivalent to what a chemicals operator already has.
What QbD does not do
QbD does not eliminate batch testing. The framework explicitly allows for - and most filings include - a mix of in-process controls and finished-product release tests. The mix is a judgement call that depends on how well the process is understood and how aggressively the manufacturer wants to claim that understanding.
QbD also does not, by itself, justify skipping chemometric model validation. A model embedded in a PAT instrument is an analytical procedure, and the validation expectations now live in ICH Q2(R2) and Q14. A QbD filing that names a PAT instrument is a filing that has to support the model behind it.
Finally, QbD is not a one-time exercise. The framework presumes that process understanding accumulates over the lifecycle, and ICH Q12 codifies the mechanisms for keeping the regulatory file in sync with that learning. A QbD-aligned filing closes one chapter; it does not close the book.
The practical takeaway for operators is straightforward. PAT instruments matter because they make a control strategy executable on the cadence the process needs. QbD is the framework that explains why a control strategy is the right object to be designing in the first place.