CMS Pub. 100-06, ch. 8, § 60.6

Design such tests as are necessary to accomplish your audit objectives. Your tests must

aid you in reaching conclusions necessary to complete the audit. Use sampling when this

would be more efficient in testing the universe of transactions, entries, or statistical data

within an area of consideration.

Sampling is the application of an audit procedure to less than 100 percent of the items

within an account balance, class of transactions, or statistics (e.g., count of

interns/residents) to evaluate some characteristic of the such balance, class, or statistics.

On the basis of facts known to you, decide if all transactions, balances, or statistics that

pertain to the issue/area being tested need to be reviewed in order to obtain sufficient

evidence. In most cases, an auditor will test at a level less than 100 percent.

There are two general sampling approaches, nonstatistical and statistical. Either

approach, when properly applied, can provide sufficient evidential data related to the

design and size of an audit sample, among other factors. A nonstatistical sample may

support acceptance of findings, but findings must be scientifically established to support

adjustments.

Some degree of uncertainty is inherent in applying audit procedures and is referred to as

ultimate risk. Ultimate risk includes uncertainties due both to sampling and other factors.

Sampling risk arises from the possibility that when a compliance or a substantive test is

restricted to a sample, the auditor's conclusions may be different had the test been applied

in the same way to all items in the account balance, class of transactions, or statistics.

If you use a sample to test certain issues scoped for audit, you must include a description

of the sampling technique, all parameters used to select the sample, and confidence level

in the audit working papers.

A. Planning Samples

Planning an audit involves a strategy for selecting appropriate sample(s). When planning

a particular sample, consider:

• The relationship of the sample to the audit objective (e.g., Medicare policies for

determining the GME and IME FTE counts of residents differ and these

differences must be considered in the decision whether it is feasible to use one

sample to test the FTE counts for both purposes);

• Preliminary estimates of materiality levels;

• The allowable risk of incorrect acceptance; and

• Characteristics of the population, i.e., the items comprising the universe.

B. Selecting a Sampling Approach

Because either nonstatistical or statistical sampling can provide sufficient evidence,

choose between them after considering their relative cost and effectiveness. Statistical

sampling helps to:

• Design an efficient sample;

• Measure the sufficiency of the evidential matter obtained; and

• Evaluate the results.

By using statistical theory, quantify sampling risk in limiting it to an acceptable level.

Statistical sampling involves additional costs of designing individual samples to meet the

statistical requirements and selecting items to be examined. Where the audit objective

would be best accomplished by stratifying the universe/population into high and low

strata (e.g., where Medicare bad debts are being tested), use your judgment in designating

the threshold for this stratification. Once determined, review all the items in the high

strata population and use statistical or nonstatistical sampling to test the low strata.

C. Sampling Risk

In performing substantive tests of details, consider:

• The risk of incorrect acceptance that the sample supports the conclusion that the

items are not materially misstated when they are; and

• The risk of incorrect rejection that the sample supports the conclusion that the

items are materially misstated when they are not.

D. Using the Test Results

If the results of testing your sample that was selected using a nonstatistical method

indicate probable errors in the universe of transactions, entries, or statistics, document

your decision to expand the sample or redesign the sample using a statistical method. If

the results of testing your sample that was selected using a statistical method indicate

probable errors in the universe, document your decision to project the error to the

universe/population.

If your adjustment pertains only to the error(s) that was identified, you must document

the reason for not considering the effect of the error(s) on the universe.