APR and PQR: The Yearly Review Nobody Reads

21 CFR 211.180(e) requires a written record reviewing each drug product at least annually, to determine whether specifications or procedures need changing.

The EU equivalent is the Product Quality Review (PQR) under EU GMP Chapter 1. Same intent, broader scope, and most companies producing for both markets write one document satisfying both.

What it has to cover

The US requirement is short. It asks for a review of representative batches, including those that failed specification, together with complaints, recalls, returned or salvaged product, and investigations.

The EU PQR is more prescriptive and effectively supersedes it for anyone doing both. It adds starting and packaging materials, in-process and finished product results, out-of-specification results and their investigations, changes, variations submitted and approved, stability results and adverse trends, returns, complaints and recalls, corrective actions, equipment and utility qualification status, and the state of technical and quality agreements.

What separates a useful one from a compliant one

A compliant APR restates a year of data. A useful one draws a conclusion.

The distinction shows in whether the document answers:

• Is the process still capable? Trends in yield, in-process results and finished product data, with capability actually calculated rather than asserted.
• Are the specifications still right? A result that sits at 95% of a limit every batch is telling you something.
• Did the deviations share causes? Twelve deviations with one root cause is a systemic finding that only appears when they are read together, which is the one thing an annual review is uniquely positioned to do.
• Did anything change without the review noticing? Suppliers, sites, methods.
• What are we doing about it? A review producing no actions in a year where something went wrong has not been read by anyone with authority.

Where it goes wrong

Written the week it is due, from data nobody has looked at since it was generated.

No trending, only listing. A table of results is not a trend. The requirement is to determine whether changes are needed, which cannot be done without looking across time.

Actions with no owner or date, which makes the next review a repeat of the last one.

Disconnected from change control. The review notices a specification is marginal and no change control follows, so the same observation appears the next year.

An inspector reading two consecutive reviews that make the same observation with no action between them has learned something about the quality system rather than about the product.