Frequently Asked Questions on Statistical Quality Control

1. What is the difference between Statistical Quality Control (SQC) and statistical process control (SPC)?

Statistical quality control (SQC) is the use of statistical tools to monitor and maintain product quality. Statistical process control (SPC) is the use of statistical tools to assess the quality of a process, typically a production operation.

2. I have several different dosage form products and each has a different SOP to follow to measure the weight variation. How can this be handled more efficiently?

XPR balances can store up to 50 different application methods directly on the balance. Once you have configured your SOP method, it can be started from a shortcut on the balance home screen. With one tap on the shortcut, the balance user has the required SOP with all the preset tolerances ready to go.

3. What is the difference between the weight variation test and the content uniformity test?

In the case of pharmaceutical dosage forms, the weight variation test (also referred to as uniformity of dosage units) is a non-destructive test that compares the individual weights of a sample of tablets with the average weight of the selected sample. The tablets are selected at random from the same production batch, weighed individually and the weight variation determined. USP requires that no more than 2 tablets are outside the specified limit and none should be over twice the limit. In general, the weight variation test is suitable for tablets, hard capsules and solids in single unit containers but not for soft capsules and certain coated tablets.

The content uniformity test is a destructive test that requires an assay to determine the quantity of active pharmaceutical ingredient in each dosage unit. The test is performed on a random selection of tablets from a batch. The mean value of the samples is compared against the reference value; the batch is accepted or rejected according to the defined allowable deviation. In general, the content uniformity test is used for inhalations, suppositories, patches and certain coated tablets.

4. Is calculation of the mean value and standard deviation sufficient?

According to the central limit theorem, a process that is subject to a large number of influences, none of which are dominant, will always show results that follow the laws of the normal distribution. The normal distribution has been found to be the best basis for the evaluation of the scatter behavior typical of filling methods. The mean and the standard deviation provide an unambiguous definition of a normal distribution. If your fill quantity tests show that the results are not normally distributed, this can mean that the filling process is dominated by one or more influences and, hence, it is likely that improvements could be made.

5. How many samples do I need to check?

The number of samples required for testing depends upon the type of sample and the applicable regulations. For example, USP states that not less than 30 dosage forms should be selected for the weight variation test. As a rule of thumb, it is better to take a small number of random samples frequently than a large number infrequently. This approach is useful in filling processes that are less stable.

6. We do our weight variation tests manually which is quite time-consuming because all data must be checked and evaluated twice. How can we make the process more efficient?

Connect an LV12 automatic feeder to your XPR analytical or precision balance. The LV12 automatically dispenses your dosage forms one by one into a container on your balance. Each weight value is captured automatically and the balance tares itself in between. Once you have set in the system in action, you are free to work on other tasks.

7. We usually do our SQC manually but sometimes the results are a bit variable. What could be wrong?

Manual process are inherently prone to human error. There may be a counting error or a transcription error that could easily skew your results. In such cases, an investigation should be undertaken to determine if an outlying result was due to an error from the operator or a fault in the product. To help avoid such challenges, a move to more automated approach would help.