In-Process Check

An in-process check is a planned verification, test, or observation performed during a manufacturing operation to confirm that equipment, materials, intermediates, subassemblies, or environmental/operational conditions meet predefined in-process specifications before production continues. These checks are part of the process control strategy, not end-product testing. They range from physical and chemical measurements (e.g., blend uniformity, viscosity, pH, tablet weight, torque, seam integrity, fill volume) to visual verifications (e.g., label/lot correctness) and equipment status confirmations (e.g., setpoint, ID, calibration).

Regulators expect appropriate in-process checks to detect process variation early, prevent defects, and generate contemporaneous evidence within batch/device history records. 21 CFR 211.110 mandates sampling and monitoring of in-process materials (e.g., blend uniformity, dissolution where applicable). 21 CFR 820.80 requires in-process acceptance activities for medical devices. 21 CFR 111.75 requires dietary supplement manufacturers to establish specifications and confirm them using appropriate tests or examinations. Food operators implement risk-based preventive controls under 21 CFR 117.130; many are operationalized as in-process checks under HACCP/HarPC plans. In MES terms, these checks are procedural steps with explicit acceptance criteria, signatures, and, when possible, automated data capture.

• Purpose: verify process health before downstream value is added
• Scope: parameters, materials, equipment states, documents/labels, environment
• Evidence: contemporaneous entries with attribution, date/time, result, status
• Action: continue, adjust, rework, hold, or reject per predefined rules

Pharmaceuticals: 21 CFR 211.110 requires control procedures to monitor output and validate performance of those manufacturing processes that may cause variability. Examples include sampling of in-process materials to ensure conformance to specifications such as identity, strength, quality, and purity. Acceptance criteria and sampling plans must be scientifically sound, with ongoing monitoring trending toward continued process verification consistent with ICH Q10’s Pharmaceutical Quality System and the lifecycle approach articulated across ICH Q8/Q9/Q10/Q12.

Medical devices: 21 CFR 820.80 mandates in-process acceptance activities with documented procedures, criteria, and records linking acceptance status to product identity. Dietary supplements: 21 CFR 111.75 requires specifications (component, in-process, and finished) and confirmation using appropriate tests or examinations. Food: 21 CFR 117.130 requires a hazard analysis and risk-based preventive controls; many controls (e.g., cook temperature, pH, metal detection) are structured as in-process verifications. Across sectors, evidence must be attributable, contemporaneous, original, accurate, and complete (ALCOA+), with audit trails and secure access per data integrity guidance and relevant computerized system controls.

Effective in-process checks emerge from risk-based process understanding. Identify critical process parameters (CPPs) that influence critical quality attributes (CQAs) and define checks with acceptance limits, frequency, and control limits aligned to development data and process capability. Where continuous monitoring is practical (e.g., continuous temperature/pressure/torque), define alarm and action limits; where discrete inspection is needed (e.g., fill volume headspace verification), establish sampling plans justified by variability and risk. Tie each check to explicit operational decision rules (continue, adjust, hold, rework), including escalation and communication pathways.

• Define specification vs. control limits vs. guard bands (avoid frequent boundary hits).
• Select measurement methods/instruments with adequate accuracy and uncertainty budgets.
• Establish statistically sound sampling plans and OCAPs (Out-of-Control Action Plans).
• Document linkages: CPP→CQA, check→acceptance criteria, alarm→interlock/hold.
• Trend and review performance; refine frequency using capability (Cp/Cpk) evidence.

Ensure procedures define who performs the check, when (entry/exit criteria of a step), how (method/SOP), what to record (raw values, pass/fail, lot/equipment IDs), and what to do upon failure. Integrate with change control so any change to limits, methods, or frequency is assessed and approved. Align with ICH Q10’s management responsibilities and knowledge management so lessons learned feed continuous improvement.

In an MES, in-process checks are structured as enforceable operation steps with required data fields, device integrations, and signatures. Authoring specifies: instrument binding (e.g., scale ID), allowable ranges, units, rounding, and exception flows. Execution should minimize manual transcription by integrating data sources (e.g., OPC/PLC tags, checkweighers, torque analyzers, vision systems), with automatic status evaluation. Where human verification is necessary (e.g., label/lot verification), enforce double-checks or electronic witnessing per risk.

Procedural controls include forced signatures, formula-locked steps, range checking, interlocks that block advancement on failure, and conditional branching to deviation workflows. Records must include lot/material IDs, equipment ID and status, timestamps, operator identity, and audit trail entries for any change or override. Exception resolution (retest, adjustment, rework, hold) should be captured with rationale and approvals to maintain traceability and data integrity.

Not all checks are 100% inspections. Where product risk and process capability permit, sampling plans can be used, but must be statistically sound and scientifically justified. For continuous variables (e.g., weight, torque), SPC is preferred: apply control charts (X̄-R, individuals, EWMA) with defined Western Electric/Nelson rules, and link out-of-control signals to OCAPs. Acceptance decisions should distinguish between control limit excursions (process instability) and specification limit failures (product nonconformance).

• Use measurement systems with acceptable Gage R&R; validate methods and instruments.
• Set sampling frequency based on risk, variability, and historical performance.
• Automate charting and rules evaluation; minimize manual calculations.
• Document scientifically valid rationale (tie to development data and ongoing CPV).

Dietary supplement regulations require scientifically valid methods (e.g., 21 CFR 111.75 in conjunction with method validity principles in the CGMP subparts), and food preventive controls must be validated or verified as appropriate. Formal SPC and trending support ICH Q10’s process performance and product quality monitoring system. Where discrete attribute checks are used (e.g., label correctness), define clear acceptance rules (zero-tolerance for critical defects) and re-inspection criteria following an exception.

In-process check records must be ALCOA+: attributable, legible, contemporaneous, original, accurate, plus complete, consistent, enduring, and available. Computerized systems used to capture these data are expected to control user access (RBAC), enforce unique electronic signatures, provide secure audit trails for create/modify/delete, time-stamp events with synchronized clocks, and protect data at rest and in transit. These expectations align with 21 CFR Part 11 (where applicable), EU GMP Annex 11, and MHRA GxP data integrity guidance.

Audit trail review should be risk-based and timely, especially for parameters that drive accept/reject decisions or trigger holds. Exception-based review can be leveraged when system controls and validation evidence demonstrate that in-spec data are reliably captured without operator bias. Batch record review must confirm that all required in-process checks were performed as specified, met acceptance criteria, and that any exceptions were properly investigated and dispositioned before release.

• Mandatory fields and device binding prevent incomplete or misattributed entries.
• Reason-for-change and justification fields discourage data manipulation.
• Electronic witnessing supports high-risk verifications without paper transcriptions.
• Automated linkages to deviations/CAPA ensure traceable escalation.

ISA‑95 clarifies the partition of responsibilities across enterprise and control levels. In-process checks are authored and executed primarily at Level 3 (MES) with data exchange to Level 2 (SCADA/DCS/PLC) and Level 4 (ERP/QMS) for materials, equipment, and quality status synchronization. ISA‑88 recipe models map checks into unit procedures, operations, and phases; equipment and control modules may automate measurement and enforce permissives/interlocks tied to acceptance criteria. Proper interface definition reduces manual transcription and strengthens data integrity.

Define semantic tags and event frames so MES can contextualize measurements to batch, operation, and equipment states. Ensure bidirectional handshakes for holds: if a check fails, MES should set product/equipment to a hold state and block further automated actions until cleared by authorized quality personnel.

Failures during in-process checks should be triaged with clear definitions: Out-of-Expectation (OOE) for data anomalies or measurement system issues; Out-of-Trend (OOT) for statistically significant drifts still within spec; Out-of-Specification (OOS) when a result violates established acceptance criteria. Each class has predefined actions: OOE drives instrument checks or repeatability checks; OOT triggers engineering/quality review and potential parameter adjustment; OOS mandates immediate hold, material segregation, and formal deviation/OOS investigation per SOP.

• Immediate actions: stop/hold, line clearance as needed, assess affected scope/time window.
• Document retest/reinspection criteria to avoid testing into compliance.
• Perform impact assessment on already-processed units (backward/forward trace).
• Escalate to CAPA when root cause indicates systemic process or method issues.

Acceptance and disposition decisions should be traceable to documented rationale, with approvals captured via electronic signatures. Where rework is allowed, define validated rework processes and additional in-process checks to confirm that quality is restored without introducing new risks. Release cannot proceed until all required checks are complete, exceptions closed, and quality has approved the disposition.

In-process checks rely on qualified instruments and validated methods appropriate to the parameter and range. Device selection should consider measurement uncertainty, environmental influences, and integration capability. Inspection, measuring, and test equipment must be calibrated at defined intervals, with status visible at the point of use; methods must be demonstrated as scientifically valid with known accuracy/precision. Where automated data flows exist, interface validation must verify data mapping, units, rounding, and boundary conditions.

• Define method SOPs with sample handling, instrument settings, and acceptance rules.
• Qualify instruments (IQ/OQ/PQ) and manage calibration with traceable standards.
• Perform Gage R&R and bias/linearity assessments for critical measurements.
• Bind equipment IDs in MES to enforce use of in-calibration devices only.

For attribute checks (e.g., visual inspection), control human factors with training, challenge sets, environmental/lighting specifications, and periodic proficiency checks. For automated systems (vision, checkweighers), maintain validated libraries/recipes and periodically challenge with known-good/known-bad standards to verify detection performance and false-positive/false-negative rates.

Computerized support for in-process checks must be validated commensurate with risk. Apply GAMP 5 (2nd ed.) principles: perform risk-based categorization, author URS focusing on data integrity and decision logic, execute installation/operational/performance qualification (IQ/OQ/PQ), and verify that workflows enforce acceptance criteria, signatures, and exception handling. Keep a configuration management baseline and traceability matrix linking requirements to test evidence; revalidate on significant changes to acceptance criteria, interfaces, or device drivers.

• Govern master data: limits, sampling frequencies, methods, instrument bindings.
• Route changes via formal change control with impact assessment across PQS.
• Periodically review process performance (CPV) and adjust frequencies/limits.
• Train personnel and maintain role-based access aligned to responsibilities.

Quality oversight reviews completeness, right-first-time metrics, exception rates, and human overrides to detect latent control gaps. Management reviews evaluate the effectiveness of in-process controls as part of the site’s Pharmaceutical Quality System (ICH Q10), including knowledge management to feed improvements to methods, automation, and training.

V5 Ultimate implements in-process checks as first-class MES objects with tight integration across eBMR/eDHR, QMS, LIMS, WMS, and equipment interfaces. Authoring defines limits, device bindings, SPC rules, and exception paths. Execution enforces mandatory data capture, unique e-signatures, and audit trails; automated device ingestion minimizes transcription error. Failures trigger immediate holds and open linked deviations/CAPAs; release is blocked until closure and quality approval. ISA‑95-aligned connectors synchronize material/equipment status and ensure that only in-calibration devices can be used.