INTRODUCTION
As indicated in the General Notices, Uniformity of Dosage Units 〈905〉 is a specification for an article of commerce and not intended for batch release. Instead, companies are encouraged to introduce batch release criteria to control the uniformity of produced material to ensure that material meeting their release criteria would pass 〈905〉 if tested once. USP is aware that USP tests are commonly used for batch release. However, the procedures and acceptance criteria in 〈905〉 refer only to the units tested, so any passing results should not be extrapolated to larger populations. The approach to release the batch (with respect to content uniformity) on the basis of passing the 〈905〉 requirement is neither appropriate nor acceptably rigorous.
This chapter presents an approach to testing for content uniformity that is focused on ensuring that an article of commerce will meet the requirements of 〈905〉 utilizing appropriate quality risk management principles. It is important to note that the batch release test and associated requirements constitute a company-specific internal criterion to ensure meeting the 〈905〉 criterion. Other possibilities are covered in the Stimuli article Batch Release Testing for Ensuring Compliance with Uniformity of Dosage Units〈905〉 (1).
BASIC PRINCIPLES
USP tests are not intended nor designed for batch release (2), though are commonly used in that fashion. If statistical testing is to be used for making batch disposition decisions, a larger test sample and/or tighter acceptance criteria than those of 〈905〉 are normally required to be able to judge whether the batch (with high likelihood) is acceptable.
The basic principles for statistical batch release testing are simple and well established:
One should have a defined quantitative quality goal (i.e., the probability that the batch, if tested, would meet the requirements of 〈905〉 is at least 95%)
A sample of results is collected from the batch being assessed; the collection of this should be organized so it provides a fair representation of the batch (see below for discussion of sampling options)
On the basis of the results from the sample, the batch is judged to fulfill the quality goal or not
All elements of the decision procedure should be determined before applying it
It should be acknowledged that all statistical decision procedures are associated with errors. There will always be some risk that a “good” batch (i.e., one complying with the quality goal) by chance fails acceptance criteria and is rejected; this is the producer’s risk, PP. Similarly, there will also be a risk that a “bad” batch (i.e., one that does not fulfill the quality goal) passes requirements and is released to the market; this is the consumer’s risk, PC.
A critical element of the decision process is to ensure that the sampling plan will produce a sample that is representative of the batch being assessed. The batch may be sampled using various sampling plans such as random, stratified, or systematic. Random sampling is common for solid dosage forms, such as tablets, after the dosage units have been thoroughly mixed (e.g., after coating). Stratified sampling is often used when the batch is partitioned into “stratum” (e.g., when parallel filling of containers is used); then units to be tested are randomly sampled within each “stratum”. Systematic sampling can be used when a drift during manufacturing cannot be ruled out; units are then collected at approximately equal intervals throughout the batch.
For the case of assessing uniformity of dosage units (UDU), it is natural that the quality goal is described in terms of the probability to pass 〈905〉 (we denote this as P905). Thus, a manufacturer could select the quality goal as "the probability that an item in commerce that is produced from the batch material will meet the requirements of 〈905〉 is greater than 95% (P905 ≥95%)". In this case the manufacturer should design their release test so that an article of commerce from a batch that is released will have at least a 95% chance of passing 〈905〉 if tested (once). As noted, due to the sampling element, there will be classification errors; a proportion (PP) of “good” batches (that should meet the goal of P905 ≥95%) will be rejected despite fulfilling the goal, and a proportion (PC) of “bad” batches (that should not meet the goal of P905 <95%) will be wrongly released. The probabilities, PP and PC, can be reduced by increasing the sample size, but it is not possible to make both probabilities zero. Also, there will always be a tradeoff between the specificity and sensitivity of this testing approach.
It should be noted that the general procedure described assumes that the properties of the batch remain unchanged during product shelf life. If there is a suspicion that the batch mean level is changing with time, the release test should be modified to take this into account.
There are numerous suggestions on how to use the information contained in a sample to develop an informed decision. The tolerance interval approach proposed here is provided as a guidance; there may be other acceptable statistical batch release tests that can be used equally well if properly designed to the same goal (3–4).
Source from USP and Please refer to USP for details:
plastic pharmaceutical packaging
Copyright © Shijiazhuang Xinfuda Medical Packaging Co., Ltd. All Rights
MAKE AN ENQUIRY
