This artcle disscuss on- how to set the limit of Out of Expectation (OOE)/ Out of trend (OOT)?

Simply, if we get tested results/ data outside of the usual trend, can be considered OOE. This is also known as Out of Trend (OOT) though OOT is more appropriate for stability data. 

OOE/OOT Limit can be set as mentioned below:

1. Based on statistical assessment: A statistical assessment should only be made where: a) The production process is stable b) There is continuous data entry c) There are at least 15 data points. If using statistical assessment, determine the process capability (PpK) then follow the guidelines mentioned below for the calculation of OOE limits.

  • If PpK >3.0,  OOE = Process mean ± 6 x process standard deviation.
  • If PpK ≥1.0 to 3.0, OOE = Process mean ± 3 x process standard deviation.
  • If PpK <1.0, OOE = Registration specification – analytical variability.

Analytical variability may be estimated from a number of different sources including method validation, method transfer, literature, and Pharmacopoeia.

Data from batches where a special cause has been identified e.g. process deviation and which are not part of the normal population may be eliminated prior to determination of the OOE limits, as these do not represent the normal production process.

2. Based on scientific judgment: Where a statistical assessment is not used to establish the OOE limits, scientific judgment from previous experience e.g. previous campaigns should be used. OOE limits should be set at a point where results become analytically significant or significant to API product quality and not only statistically significant. This may take into account the expected analytical variability, historical review of results, and the likelihood of failure to meet requirements.

3. New Processes, Manual Processes, and Processes where historical data is not readily available: Where historical data is not available to implement OOE limits based on statistical evaluation or scientific judgment, an OOE limit of 80% of the specification should be assigned (or tighter based on the scientific judgment). When sufficient data (minimum 15 batches) is available, evaluation and implementation of OOE limits must be redefined. 

Resource Person: M. Raihan Chowdhury