Sampling Plans

• Instructions
• Tables

Sampling Plan Instructions

1. Select the table based upon how sure you want to be about what is observed. For example, if you are reviewing Device History Records of a life supporting device, you may choose to use Table 2 (99% Confidence). You may choose to use Table 1 (95% Confidence) for the review of Device History Records regarding a device with lower risk.
2. Select a sample size. If the population of records to be sampled is small (approximately thirty or less), you may choose to review all of the records.
3. Review the sample of records selected. You can terminate your review of the entire sample if you observe objectionable conditions beyond the number stated in the column header*. However, if you do not review all of the records in the sample, you may not report additional infor-mation that could be useful in further understanding the potential prevalence of the objectionable condition observed, or you may not recognize whether other objectionable conditions exist.

   (* If you choose to terminate your review prior to completing the review of the entire sample, in addition to the information contained in instruction 4, report in the Establishment Inspection Report how many records were reviewed prior to your termination of the review.)

Note: Factors to consider when selecting a sampling table and sampling size may include the risk of the device being inspected or the records being sampled, and the amount of time you have located to this portion of the inspection.

4. When objectionable conditions are observed based upon samples chosen using these tables, report in the Establishment Inspection Report: (a) the total number of records included in the population from which the sample was chosen; (b) the table used to select your sample; (c) the row used to select your sample; and, (d) the sample size selected*.

   (* The information requested in instruction 4 must be reported whenever an Official Action Indicated (OAI) endorsement is considered. Reporting this information may not be necessary when Voluntary Action or No Action is indicated. However, caution is advised when using this reporting discretion because Voluntary Action Indicated endorsements are sometimes elevated to Official Action Indicated.)

NOTE:

A. There are no "acceptable" violations of the Quality System Regulation. All Quality System Regulation violations encountered must be handled appropriately according to current FDA policies and procedures. When using the "1 out of:" and "2 out of:" columns, it does not mean no more than that number of Quality System Regulation violations per the appropriate sample size is acceptable. It will only give you an initial understanding of how prevalent the problem may be.

B. When at all possible, all samples should be chosen at random.

Table 1
Binomial Staged Sampling Plans
Binomial Confidence Levels

Confidence Limit .95 <>		0 out of:	1 out of:	2 out of:
A	.30 ucl*	11	17	22
B	.25 ucl	13	20	27
C	.20 ucl	17	26	34
D	.15 ucl	23	35	46
E	.10 ucl	35	52	72
F	.05 ucl	72	115	157

Table 2
Binomial Staged Sampling Plans
Binomial Confidence Levels

Confidence Limit .99 <>		0 out of:	1 out of:	2 out of:
A	.30 ucl*	15	22	27
B	.25 ucl	19	27	34
C	.20 ucl	24	34	43
D	.15 ucl	35	47	59
E	.10 ucl	51	73	90
F	.05 ucl	107	161	190

*ucl = Upper Confidence Level

CRC Handbook of Probability and Statistics: Second Edition

Binomial Sampling may be used when trying to make a decision about an endpoint that only has two potential outcomes (e.g., The device history record is compliant or the device history record is noncompliant).