Analytical quality
How to Compare Results Between Batches
Two batches can meet a specification while showing differences that deserve review. They can also yield different figures without demonstrating a material change. A useful comparison identifies the evaluated property, available history and evidence supporting interpretation of the difference.
Source editorial review:
Meeting a criterion does not demonstrate equality
ICH Q5E states that routine specification tests are generally insufficient to fully evaluate a manufacturing change. It also considers results within acceptance criteria that depart from historical trends.
The guideline concerns biotechnological and biological products. Its distinction is useful when reading data: passing a limit and comparing batches are different conclusions. The guideline should not be presented as certification of a particular material. ICH Q5E.
Compare the profile underlying the result
ICH Q6B describes characterization of heterogeneity and batch consistency in protein products. A profile can contain information that does not fit into a total figure.
Hypothetical example: two batches show the same overall percentage but differ in the distribution of secondary components. The percentage does not identify which components changed. The comparison should preserve profiles and available assignments without inventing identities for unknown signals. ICH Q6B.
Time order provides information
NIST describes control charts plotting a characteristic against time or sample number. A result's position is interpreted relative to process behavior, not only through an isolated comparison between two figures.
When reading a history, preserve batch order and the variable's definition. Sorting from lowest to highest may serve another question, but loses the sequence needed to examine temporal changes. NIST: control charts.
A control signal starts an investigation
In NIST's explanation, control limits entail a risk of flagging variation that could be due to chance. A point outside the expected pattern therefore does not automatically identify its cause.
The report must distinguish the statistical signal from the confirmed explanation. Immediately attributing it to degradation, contamination or a manufacturing change exceeds what the chart demonstrates. The comparison guides investigation; the cause needs additional evidence.
Document differences and the conclusion's scope
ICH Q5E calls for evaluating profile differences and recognizes that comparability does not require absolute identity of all attributes. That evaluation depends on knowledge of the product and the studied change.
As a reading framework, separate similarities, observed differences and unresolved questions. ICH Q6B also distinguishes quantity, purity and activity among characterization attributes: agreement in one does not establish agreement in all.
A useful conclusion can be limited to the properties actually compared. If only two summarized certificates exist, questions those documents cannot answer must remain open.
Questions and answers
Are two compliant batches necessarily equivalent?
No. Compliance and comparability require interpretation of different questions and datasets.
Does an out-of-control point identify an impurity?
No. It flags behavior that warrants investigation; it does not determine the identity or origin of a difference.