ICH Q13: Continuous Manufacturing of Drug Substances and Drug Products

ICH Q13 is the International Council for Harmonisation guideline on continuous manufacturing of drug substances and drug products. It provides a framework for connecting process understanding, material movement and evidence of product quality.

For manufacturing teams, the practical questions are: which material belongs to a batch; how disturbances propagate through connected equipment; when output can be collected; and which records support disposition? These questions connect batch identification to the evidence reviewed at the QA disposition step. Continuous manufacturing is therefore a material-flow and evidence problem, not merely a decision to run equipment for longer.

Keep four categories distinct: what Q13 recommends or describes, what applicable regional law requires, what operating practices a manufacturer proposes, and what software supports. The sections below follow those boundaries from process scope and submissions through control, material tracking, batch reconciliation and disposition.

ICH Q13 section 1.2 describes CM as “the continuous feeding of input materials into, the transformation of in-process materials within, and the concomitant removal of output materials from a manufacturing process.” The guideline applies to CM of drug substances and drug products for chemical entities and therapeutic proteins, to new products and to conversion of existing batch processes, and its principles may also apply to other biological/biotechnological entities. Although the description can fit a single unit operation, Q13 focuses on integrated systems “in which two or more unit operations are directly connected”, because a change in one operation can affect upstream and downstream operations and output quality.

ICH Q13, Section 1.2 covers continuous manufacturing of drug substances and drug products for chemical entities and therapeutic proteins, including new products and conversion of existing batch processes. Assess three layers separately: guidance recommendations, applicable law and product authorisation conditions.

For supplements, conventional foods, medical devices and non-pharmaceutical chemicals, check the current Q13 scope and the legislation applicable to the product before adopting a pharmaceutical framework. Start with product classification when identifying the relevant GMP framework.

A mixed architecture is illustrated in ICH Q13, Annex II, Section 1: continuous feeding, blending and tablet compression lead to batch film coating. This example does not make every operation continuous.

As a practical site-scoping exercise, identify drug substance versus drug product, included unit operations, intermediate storage, downstream batch steps and interfaces with packaging. Mark these boundaries on the process map; use bin-level lot genealogy to examine material handoffs through intermediate storage.

A continuous manufacturing dossier should connect how the line operates with the evidence supporting its controls. Treat the Q13 recommendations below as guidance, not as a substitute for the applicable regional submission requirements. Keep the operating description, development rationale and proposed validation approach consistent with one another.

Map operating knowledge to the CTD

ICH Q13, Section 4.1 recommends a sequential narrative of the manufacturing process and process controls in CTD sections 3.2.S.2.2 and 3.2.P.3.3, supported by pharmaceutical development data in 3.2.S.2.6 or 3.2.P.2. For applicable CM processes, that description should also include:

• Summaries of start-up, shutdown, pause and restart procedures explaining how the integrated process functions.
• The material collection and diversion strategy, including criteria for collecting product during routine manufacturing.

As a practical dossier check, trace each collection criterion back to its development rationale and the relevant element of the control strategy. A procedure summary explains what operators do; the supporting development account explains why the proposed conditions are suitable. Cross-references make that relationship easier to review without duplicating entire studies.

Explain validation and plan lifecycle changes

ICH Q13, Section 4.7 describes traditional validation using a fixed number of validation batches and continuous process verification as possible process validation approaches. For continuous process verification, Q13 recommends a dossier justification based on product and process understanding, system design and the overall control strategy. Choosing that approach therefore needs an explanation, not merely a statement that continuous monitoring is available.

For conversion of an approved batch process to CM, the Q13 conversion discussion recommends seeking regulatory approval before implementation. It also identifies authority advice as a way to clarify expectations for the strategy and data package. Check the applicable regional change route for the product and proposed conversion; do not infer a filing category or implementation timetable from Q13 alone.

Building on the Annex II example above, consider a proposed direct-compression line: raw materials enter through feeders, pass through a continuous blender and tablet press, and reach collection containers before downstream batch coating. The sequence below is an engineering planning approach, not a prescribed universal implementation method.

Engineer feeder replenishment, transfer paths, surge capacity, sampling positions, container changeovers and coating capacity explicitly. Draw the material path separately from the measurement path—where sensors and samples observe it—and the decision path: which controller or person acts on those observations.

• Start-up: confirm equipment readiness, locate initial material and identify its collection destination.
• Normal production: track equipment condition, material movement and the active collection container.
• Planned pause: identify stopped equipment, material held inside it and whether collection remains open.
• Restart: assess equipment readiness and held material before resuming movement and selecting collection destinations.
• Shutdown: identify equipment still clearing, residual material locations and final container closure.

Assess output changes separately: longer run time, changed flow rate and changed equipment capacity. For each, ask whether transport behaviour changes, whether downstream operations can absorb output through finite capacity scheduling, and whether existing evidence remains representative. Do not assume identical regulatory treatment.

Seven-step operational workflow

1. Characterisation: study powder behaviour, feeding consistency and transport through connected equipment.
2. System design: specify interfaces, capacities, sampling access and container handling.
3. Challenge studies: test replenishment, interruptions and recovery; connect equipment testing to operational qualification.
4. Boundary definition: document operating ranges, material-routing boundaries and batch boundaries.
5. Validation: demonstrate performance under the proposed operating conditions.
6. Routine operation: assign transition checks, monitoring responsibilities and escalation ownership.
7. Change assessment: evaluate altered duration, flow or equipment against established knowledge.

Q13 section 3.1.1 uses the ICH Q10 meaning of state of control: “a condition that provides assurance of continued process performance and product quality.” It is not the same as steady state. Q13 notes that a state of control can be demonstrated when parameters such as process parameters and quality attributes are changing within specified ranges, while the process is not necessarily in a steady state. Control-strategy elements monitor that condition, and sites need mechanisms to detect when parameters drift or trend within the specified range.

A continuous-manufacturing control strategy should connect incoming material, process behaviour and output quality—not merely keep instrument traces level. As an engineering approach, start with each relevant critical quality attribute, identify what could affect it, and determine which measurements and responses provide meaningful protection.

ICH Q8(R2), Section 2.5 illustrates control-strategy elements: input-material controls based on effects on processability or quality; product specifications; controls on unit operations affecting downstream processing or quality; and monitoring to verify multivariate prediction models. These are guidance examples, not a universal equipment checklist. In practical design, a critical material attribute assessment connects incoming variability to the operation that must accommodate or detect it.

ICH Q13, Section 5 defines steady state narrowly: “A stable condition that does not change over time.” A stable signal can nevertheless be biased, insensitive to a relevant defect, or consistently outside an acceptable range. For engineering assessment, ask whether the combined evidence supports intended process performance and output quality. Signal flatness alone does not answer that question.

For continuous process verification, ICH Q13, Section 4.7 describes real-time data demonstrating maintenance of a state of control and desired output quality for the run-time duration. The practical process validation question is therefore whether evidence covers the operating period, rather than only a favourable snapshot. Review trends together with material results, measurement reliability and model validity.

The following design-risk examples are engineering proposals, not prescribed Q13 controls. Select signals, limits and responses for the actual product, equipment and validated operating conditions.

For each proposed control, document what it detects, what it cannot detect, who evaluates an exception and what evidence supports recovery. This makes the strategy testable: a passing sensor value or model prediction is useful only when its measurement basis remains credible and its relationship to quality remains applicable.

ICH Q13, Section 5 defines residence time distribution (RTD) as “A measure of the range of residence times experienced by material passing through a specific process environment/vessel/unit operation.” A single average residence time summarises that distribution; it does not describe how early some material arrives or how long later-arriving material remains in transit. That spread matters when connecting an upstream disturbance to downstream material.

Build a process-specific transport model

As an engineering approach, build an evidence chain rather than treating a nominal delay as sufficient:

1. Use an appropriate tracer or material-response study that represents transport through the relevant process segment.
2. Time-align the input event and downstream measurements, accounting for clock offsets, sampling intervals and sensor delay.
3. Define the operating-condition envelope: flow rates, equipment configuration and material properties represented by the study.
4. Assess early and late tails, including whether measurement sensitivity could hide low-level carryover.
5. Evaluate model uncertainty and justify the timing margins used for the intended diversion decision.

Do not assume one RTD measurement remains applicable after changing throughput or equipment connections. Linking the disturbance and predicted exposure interval in an event frame helps preserve their relationship; connecting that interval to material genealogy identifies the downstream material involved.

Worked case: propagate the whole interval

Assume constant outlet flow of 60 kg/h, synchronised clocks, no recirculation and an assessed 3–7 minute transport window suitable for this hypothetical calculation. An input disturbance lasts from minute 20 to minute 22. Apply an additional one-minute guard on each side.

• Earliest affected arrival: 20 + 3 = minute 23.
• Latest affected arrival: 22 + 7 = minute 29.
• Guarded interval: minute 22 through minute 30.
• Duration: 30 − 22 = 8 minutes.
• Diverted mass: 60 kg/h × 8/60 h = 8 kg.

A mean-only five-minute shift would select minutes 25–27. It would miss portions of the assumed affected interval: minutes 23–25 and 27–29. The broader window therefore changes both routing time and the mass assigned to diversion.

The 3–7 minute window and one-minute guards are case assumptions, not Q13 defaults or evidence that real RTD tails end sharply. Actual diversion timing needs process-specific transport and uncertainty evidence, together with an allowance for routing-system response where relevant.

In ICH Q13 section 2.2, the ICH Q7 batch definition applies to all modes of continuous manufacturing (CM), for both drug substances and drug products. Q13 then says batch size can be defined by one of three bases: quantity of output material, quantity of input material, or run time at a defined mass flow rate. Other approaches are possible if scientifically justified, and a batch size can also be defined as a range, for example by setting a minimum and maximum run time. In practice, the chosen basis becomes the boundary used for traceability and disposition records.

• Batch identity identifies the material quantity subject to a defined disposition decision.
• Run identity identifies an operating interval; it should not be assumed to equal one batch.
• Production campaign groups related manufacturing activity and may contain multiple runs or batches.
• Collection container identifies a physical receptacle. Its contents need an explicit link to batch identity rather than becoming a batch solely because they share a container.

ICH Q13, Section 5 defines run time as “The time interval used to produce a quantity of output material.” The guidance’s Annex II, Section 1 illustrates defining a batch-size range using run time at a predefined mass flow rate. Treat this as a specific illustration, not the only permitted batch-definition approach.

Worked case 2: reconcile output before disposition

Assume a six-hour accounting window, measured outlet flow of 100 kg/h throughout, 20 kg diverted, and no unmeasured losses or unaccounted inventory. Gross output is 100 kg/h × 6 h = 600 kg. Retained output is 600 − 20 = 580 kg.

Here B02 and R02 are illustrative identifiers; the boundary fields explicitly describe the accounting window. The declared batch-size convention must identify whether it describes gross output, retained output or another defined quantity. Preserve both quantities in net-versus-gross reconciliation. The calculation establishes mass accounting, not acceptability: retained material still needs human review and a documented quality disposition decision.

Diversion means isolating and separating material from the manufacturing product stream, as defined in ICH Q13, Section 5. Operationally, distinguish issuing a diversion command from confirming that the affected material reached the intended destination.

As guidance, ICH Q13, Section 4.8 recommends establishing diversion procedures, when required, under the pharmaceutical quality system (PQS). It also recommends maintaining an overarching plan or decision tree under the PQS describing disturbance management for different types of diversion.

Proposed recovery sequence

1. Detect and timestamp the event; identify the potentially affected material interval using the RTD approach above.
2. Segregate that material and confirm physical routing and collection destination. Link its identity to an electronic hold tag or equivalent controlled record.
3. Investigate the disturbance and establish documented recovery criteria before resuming product collection.
4. Reconcile collected quantities with event intervals and document the disposition rationale for authorised review.

The following are proposed execution controls for site-specific risk assessment, not universal equipment requirements or automatic disposition rules.

Testing is not release authority

ICH Q8(R2), Section 2.5 explains that real-time release testing does not replace the review and quality-control steps called for under GMP to release a batch. Keep approval and batch release with authorised humans: neither a passing sensor result nor a software status grants release. Present the diversion history, investigation, recovery evidence and quantity reconciliation together so reviewers can assess the retained material in context.

These concepts answer different questions. ICH Q13, Section 3.3 describes frequent monitoring and control using process parameters, PAT tools, soft sensors and process models. The comparison below separates guidance frameworks, analytical tools, validation approaches and evidence organisation.

For detailed Q9–Q12, Q14, GMP or standards obligations, consult the current text and applicable regional scope rather than attributing additional requirements to Q13.

A proposed vendor-neutral evidence model uses five entities: run, material interval, observation, diversion event and batch disposition. This is a design recommendation, not an existing application database schema. A run has material intervals and observations; diversion events identify routing changes and link to affected intervals; a batch disposition references the included intervals and records the authorised human decision.

Assign integration responsibilities explicitly: historians preserve process observations; automation systems record routing events and recipe versions; laboratory systems link samples and results to material intervals; manufacturing records assemble the review context. For MES–historian integration, agree stable identifiers, event-time timestamps, units, signal-quality flags and controlled model or recipe versions across interfaces. Assign ownership for identifier mapping and clock alignment.

Event time means when the measurement or action occurred; arrival time means when the receiving system obtained its record. Buffering and retransmission can separate them. Preserve both, detect duplicates using source identifiers and sequence numbers, and flag conflicting retransmissions. Before record closure, explicitly check expected streams, sequence continuity, interval coverage and unresolved gaps. Absence of an alarm is not proof of acceptable production.

Worked case 3: a telemetry gap

Assume constant outlet flow of 90 kg/h throughout a 40-second telemetry gap. Gap-period output = 90 × 40/3600 = 1 kg. This is not an automatically justified diversion quantity. Apply the RTD treatment above when assessing the potentially affected interval.

Preserve available records, identify linked material intervals, assess independent measurements and document remaining uncertainty. Link the gap and recovery actions in the process event log, then route the exception through the established disposition process rather than silently closing the record.

Use this proposed multidisciplinary checklist alongside the explanations above, not as a compliance certificate. For every gate, record the named process boundary, documented operating assumptions, versioned acceptance criteria, a challenged exception scenario, a reconciled material record and an assigned decision owner. Mark unresolved evidence gaps explicitly.

Assign named owners within these functions. Resolve open issues before the associated operating decision; retain human responsibility for approval and batch release.