When you gather several breath alcohol test results and calculate an average, you must first check what each record means before looking at the size of the numbers. Test screens and data transmission records may hold not only concentration values but also detection results such as pass or fail, statuses indicating that a test did not finish, unit information, and whether data was transmitted. Which of these counts as a "validity indicator" differs by device and data format, so you must not interpret the meaning of screen messages or status values on your own.
A quantitative test here means a test that presents the alcohol concentration in exhaled breath as a number with a unit. A screening test, by contrast, may show the result as a category, such as whether a set threshold was exceeded. OIML R 126-1:2021 distinguishes quantitative evidential breath analyzers from preliminary screening devices. This distinction does not govern every device display or operating procedure in Korea, but it is a useful reference for the judgment that quantitative values and screening categories should not be treated as the same data before averaging.
For validity indicators, first confirm what they mean for each product
A validity indicator is information that tells you the processing status of a record, such as a status message, symbol, or data field attached to a result. But even when terms like "error," "incomplete," "retry," or "out of range" appear, what each one means can vary according to the device's official user manual and data definitions. So rather than immediately translating a particular indicator into a breath problem, an analysis failure, or an invalid value, keep the original message and status value and check its meaning against the documentation.
The device screen is not the only place to check. Data transmitted to management software or a server may contain result types, units, and transmission identifiers that do not appear on screen. Conversely, if the transmitted data has no status description, it is better to classify that record separately than to guess the missing information and decide whether to include it in the average.
Average candidates must be quantitative results of the same kind
Candidates for the average are quantitative values expressed in the same unit. Screening results that convey only a category, such as PASS or FAIL, are not concentration values, so do not mix them into the numerator or denominator of a quantitative average. If you convert PASS to 0 or assign an arbitrary number to FAIL, you create values the device never provided and distort the average.
Even when a displayed status exists alongside a quantitative value, you cannot automatically include or exclude it. For example, if a device transmits a number even in a particular status, whether that number can be aggregated must be decided by reviewing the official definition of that status together with the purpose of the aggregation. If you temporarily leave records whose status meaning you could not confirm out of the average calculation and keep their count and original indicator, you can correct the rule later.
| What to check in the record | Handling in the quantitative average | What to note in the report |
|---|---|---|
| Records with a confirmed concentration value and unit | Classify as average candidates within the same unit | Number of quantitative results included |
| Records with only a screening category such as pass or fail | Keep separate from the quantitative average | Number of records per category |
| Records whose status meaning could not be confirmed | Hold out of the average until interpreted | Original status indicator and reason for holding |
| Records confirmed as repeated transmissions of the same result | Count only one test result in the calculation | Number classified as duplicate transmissions |
Holding and excluding in this table do not mean deleting records. You need to keep the reason each was left out of the average so that the initial judgment can be re-checked later when reviewing the device manual, data format, and transmission history. In particular, numbers missing a unit may be in a different unit even if their digits look similar on screen, so it is appropriate not to combine them with other values until the unit is settled.
Do not lump status indicators together with transmission problems
The fact that a result is missing from the management system is not enough to conclude that the test failed. The test may have been done but the data not delivered during communication or storage, or the test may never have started. So rather than converting data that was not received into "zero concentration" or "invalid test," you need to record the status of the test result and the data receipt status separately.
Duplicates also come in two kinds that must be distinguished. If the same test result came in several times because of communication retransmission, it should be counted only once, but if the same person actually tested again, it may be a separate test record. Do not judge something a duplicate just because the numbers are the same; base the judgment on the test time and the original result identifier provided by the device or management system.
Apply the classification order in a hypothetical case
For example, suppose 12 records were received from one shift. On inspection, 8 are quantitative values with confirmed units, 1 is a screening result, 2 are records whose status meaning has not yet been confirmed, and 1 is a retransmission with the same original identifier as an earlier quantitative result. If you use all 12 as the denominator of the average, or convert the screening result and unconfirmed records to 0, you get not an average of quantitative values but a value that mixes different kinds of records.
In the interpretation step, first link the retransmitted record to the original result and treat them as one test. Next, check whether the 8 quantitative values share the same unit, and keep only the confirmed values as average candidates. A possible action is to present the 1 screening result, the 2 records of unconfirmed meaning, and the 1 duplicate transmission each as a separate count.
The completion criterion in this case is not obtaining a single average value. The total number of records received, how duplicates were handled, the number of quantitative results included in the average, and the original indicators and reasons for records not included must all be presented together. Only then can you trace which records need to be re-judged once the meaning of the status codes has been confirmed.
In the report, state what the average covers and its denominator
In the report, wording that shows what is covered, such as "average of quantitative results with confirmed units," is more appropriate than wording with an unclear scope, such as "average of all results." If you write the total number of records received, the number after removing duplicates, and the number of quantitative results included in the average alongside the average value, readers can tell what data the number came from. Present screening results and records of unconfirmed status meaning as separate counts so that no information disappears from the average.
If you have changed the aggregation rule, rather than immediately comparing whether the average is higher or lower than in previous periods, first align which result types and status indicators were included in each period. Calculating the average is the last step. Only after sorting out whether results are quantitative, whether the units match, whether the meaning of status indicators has been confirmed, and whether identical results were transmitted repeatedly does the scope that the average can explain become clear.
