What is the difference between prospective, concurrent and retrospective validation?

1What the three approaches are.

All three serve the aim PIC/S sets for process validation, to show that control strategies are robust and “manufacturing processes are reproducible”.¹ They differ in when the evidence is gathered relative to selling the batches. EU GMP Annex 15 defines prospective validation as “validation carried out before routine production of products intended for sale”, and concurrent validation as validation “carried out in exceptional circumstances, justified on the basis of significant patient benefit, where the validation protocol is executed concurrently with commercialisation of the validation batches”.² Retrospective validation draws its evidence from batches already made: ICH Q7 describes it for “well established processes” and lets retained samples be tested “to obtain data to retrospectively validate the process”.³

ICH Q7 states the order of preference plainly: “Prospective validation is the preferred approach, but there are exceptions where the other approaches can be used.”³

2Prospective validation is the default.

Annex 15 says processes developed by the traditional approach “should undergo a prospective validation programme, wherever possible, prior to certification of the product”.² PIC/S PI 006-4, its recommendation on qualification and validation in force from 1 October 2026, says the initial commercial-scale study would normally be completed before batches intended for sale are made, and that batches from a new or modified process should usually reach the market only once “both within-batch and between-batch variability” have been shown to meet pre-defined limits.¹

Before the first validation batch, PIC/S lists prerequisites: the critical quality attributes and process parameters known at the site, development and transfer reports available, qualified facilities and equipment, validated test methods, qualified suppliers, trained staff, a risk assessment of the process and its control strategy, and an approved protocol.¹ On numbers, Annex 15 says each manufacturer “must determine and justify” how many batches it needs, while “a minimum of three consecutive batches” under routine conditions is generally considered acceptable.² ICH Q7 uses three consecutive successful production batches as a guide for prospective and concurrent validation of API processes.³

FDA’s 2011 guidance arranges validation in three stages (process design, process qualification and continued process verification) rather than these three names, but makes the same point: successful completion of Stage 2 “is necessary before commercial distribution”.⁴

3Concurrent validation is an exception that needs a reason.

Annex 15 allows concurrent validation “in exceptional circumstances, where there is a strong benefit-risk ratio for the patient”. The decision “must be justified, documented in the VMP for visibility and approved by authorised personnel”, and there must be enough data to conclude that each batch is uniform and meets the acceptance criteria, available to the Qualified Person before the batch is certified.²

PIC/S adds that concurrent validation “does not negate the need for prospective validation to be performed”. The risk lies in releasing batches before the variability between batches is understood. The decision is taken before the work starts and recorded in the validation master plan and the protocol, and the justification must show that the benefit to patients, such as keeping supply of a product with no alternative, “significantly outweighs” the risk of incomplete validation knowledge.¹ PIC/S lists what makes it more likely: a therapeutically essential product that can’t easily be replaced, validation by a continuous verification approach, knowledge from a similar validated product, and a change with a foreseeable limited impact on quality, such as a different tablet shape.¹

ICH Q7 allows concurrent validation of an API process when replicate batch data are unavailable because few batches have been made, batches are made infrequently, or a validated process has been modified. Batches can be released before the study ends “based on thorough monitoring and testing”.³ It also calls concurrent validation “often the appropriate validation approach for rework procedures”.³

FDA calls the equivalent “concurrent release” of PPQ batches and “expects that concurrent release will be used rarely”, giving examples such as orphan drugs, radiopharmaceuticals with short half-lives and medically necessary drugs made in coordination with FDA to relieve a short supply. The rationale goes in the PPQ protocol, each lot released must meet all CGMPs and the protocol’s release criteria, and distributed batches need “careful oversight”, with complaints and defect reports assessed rapidly.⁴

4Retrospective validation is no longer accepted for new processes.

Annex 15, in operation since 1 October 2015, says: “Retrospective validation is no longer considered an acceptable approach.”² PIC/S PI 006-4 says the same, applies its principles to APIs and intermediates as well as finished dosage forms, and expects legacy processes validated to earlier standards to “meet current process validation standards”, including ongoing process verification with additional testing where a gap analysis and risk assessment call for it.¹

ICH Q7, the GMP guide for APIs, still describes retrospective validation as an exception for processes used without significant changes. It may be used where critical attributes and parameters are identified, in-process controls are established, there have been no significant failures other than operator error or unrelated equipment failure, and impurity profiles are established. The batches reviewed should be representative of all batches made in the period, including failures, and generally ten to thirty consecutive batches are examined.³ ICH’s Q7 questions and answers of June 2015 narrow this: prospective validation “is normally expected for processes introduced since the publication of ICH Q7”, and retrospective validation remains an exception for existing, well-established products made before ICH Q7 was implemented.⁵

FDA’s 2011 guidance does not describe retrospective validation. For legacy products it says implementation “would likely begin with the activities described in Stage 3”, continued process verification.⁴

5Validation continues after the first batches.

Whichever approach a process started with, the regulators expect monitoring afterwards. Annex 15 asks manufacturers to monitor product quality throughout the lifecycle under ongoing process verification, conducted under an approved protocol and reported, with statistical tools where appropriate.² PIC/S puts it directly: “Process validation is not a one time activity.”¹ FDA’s Stage 3 gives “ongoing assurance” during routine production that the process “remains in a state of control”.⁴

6How it runs in Seal.

In Seal, a PPQ protocol runs on the same platform as manufacturing, testing and quality. The process validation blueprint keeps the validation basis, protocol, execution records, reviewed results and exceptions as connected records, so each requirement is reviewed against its full evidence set, including missed steps and unsuccessful outcomes. Neil, Seal’s AI agent, prepares protocols and assembles evidence for review; authorised reviewers decide the conclusion.

Reviewed controls and residual risks carry into continued process verification, where trends stay connected to the batches, process versions, methods and materials behind them.