Why Quantification Limits Need Their Own Metadata
A concentration result is not fully interpretable from its number and unit alone. Whether a result is quantifiable depends on the assay and its validated range. A value of 0.015 mg/L, for example, may be within range for one assay and below the lower quantification limit for another. If the wrong limits are attached, a record can appear valid when it is not, or a valid record can be flagged incorrectly.
This work is part of making assay records consistent and reviewable. The assay harmonization step establishes which result, analyte, matrix, and assay information belong together. Limit metadata then supplies the assay-specific boundaries needed to check the result and any reported censoring status. Those boundaries do not, by themselves, determine how a censored observation will later be handled in a model.
The limit of detection is not interchangeable with the LLOQ. An analyte can be detected without being quantified with the required assay performance. Likewise, a result outside the quantification range is not automatically missing or equal to a limit. The assay report, data dictionary, and documented study rules determine what the result field and censoring code represent.
Representing the Limits and Selecting the Applicable Record
A usable metadata record should identify the assay context precisely enough to select one applicable pair of limits. Common identifying fields include analyte, matrix, assay or method identifier, laboratory, and assay version. The exact fields depend on how the source records distinguish methods. If two different methods are both called “Method A,” a label alone is not a reliable key.
Store the LLOQ and ULOQ with their units, the effective start and end dates, and the date field used to decide applicability. That date might be the analysis date, run date, or another date specified in the assay documentation. Do not assume that the sample-collection date selects the limits: it may be the right date for a particular study, but the rule must be documented. Define the interval convention as well. A useful convention is an inclusive start and exclusive end, so that a record is applicable when the selected date is on or after the start and before the end.
| Metadata field | Purpose | Check |
|---|---|---|
| Assay identity | Links a result to the correct method and context | Required identity fields are present and match unambiguously |
| LLOQ and ULOQ | Define the assay's quantification range | Both are numeric, positive where required, and LLOQ is less than ULOQ |
| Limit unit | States the scale of the boundary values | Compatible with the result unit or convertible using a documented factor |
| Effective interval | Determines when the limits apply | Start and end are valid and intervals for the same assay identity do not overlap |
| Date basis | Defines which result date selects the interval | Matches the documented assay rule and the available date fields |
These checks rely on several assumptions: the assay identity fields are accurate; the result and limits refer to the same analyte and matrix; units have known meanings; and the effective-date rule is available and consistently applied. If any assumption is uncertain, the correct response is to investigate the source or preserve the uncertainty, not to choose a convenient limit record.
Comparing Results With Limits
Before comparing a numeric result with the limits, express the result and both limits in the same unit. Under an inclusive-boundary convention, a quantified numeric result \(y\) is within range when \(L \leq y \leq U\), where \(L\) and \(U\) are the applicable LLOQ and ULOQ. A result below the range satisfies \(y < L\), and one above it satisfies \(y > U\). The assay documentation takes precedence if it specifies a different boundary convention.
The comparison is a consistency check, not permission to reinterpret the source result. A record labelled “below LLOQ” with a numeric placeholder equal to the LLOQ may be consistent if the source system uses that placeholder to encode the censoring boundary. It is not necessarily a measured concentration equal to the LLOQ. Preserve the original value and status, and distinguish a documented placeholder convention from an unexplained contradiction.
Worked Example: Converting Units Before Checking Range
Suppose a hypothetical assay record has LLOQ 0.020 mg/L and ULOQ 2.00 mg/L. The assay report gives a result of 35 ng/mL with status “quantified.” The assay documentation specifies that equality at either limit is within range. Is the status consistent?
Step 1: Put values on the same scale. Since 1 mg/L equals 1,000 ng/mL, the limits convert as follows:
Step 2: Check the range. The result is 35 ng/mL, and \(20 \leq 35 \leq 2{,}000\). It is within the quantification range, so “quantified” is consistent with these limits. Comparing 35 directly with 0.020 and 2.00, without converting units, would produce a meaningless check.
Effective Dates and Changes in Assay Limits
A method or validation change can alter the quantification range. Therefore, a metadata lookup must use both assay identity and the documented date basis. For each identity, the selected date should map to exactly one applicable limit record. No matching record means the limits are unknown for that result; multiple matching records indicate ambiguous metadata. Neither case should be resolved by selecting the first row returned by a software join.
For intervals defined as inclusive at the start and exclusive at the end, adjacent records can meet without overlapping. If one version is effective from 1 January through 1 July, represented as \([1\text{ January},1\text{ July})\), and the next begins 1 July, a result on 1 July maps only to the second version. This convention is useful only if it agrees with the source documentation. When source dates lack a time of day or use different time zones, document how dates are interpreted before assigning a boundary record.
Worked Example: Selecting Limits Across a Method Change
A hypothetical assay uses the analysis date as its date basis. Version A has limits 0.010–1.50 mg/L and is effective from 1 January through 30 June, inclusive. Version B has limits 0.025–2.00 mg/L and is effective from 1 July onward. The dataset uses dates, with the first interval represented as ending immediately before 1 July.
A sample analysed on 30 June has result 0.018 mg/L and status “quantified.” It maps to Version A, and \(0.010 \leq 0.018 \leq 1.50\), so the status is consistent.
A sample analysed on 1 July has result 0.018 mg/L and status “below LLOQ.” It maps to Version B, not Version A. Since \(0.018 < 0.025\), the censoring status agrees with the applicable limits. If the record were instead labelled “quantified,” the status would conflict with Version B under the stated boundary convention. Using Version A merely because its lower limit makes the result quantifiable would apply the wrong effective interval.
Flagging Conflicts Without Overwriting Source Data
A conflict exists when the documented meaning of the result and status cannot both agree with the applicable limits. For an ordinary numeric measurement, “quantified” paired with a value below the LLOQ or above the ULOQ is a candidate conflict. “Below LLOQ” paired with a value strictly within range is also a candidate conflict, unless the value is a documented censoring placeholder. A value above the ULOQ marked “above ULOQ” is consistent. A missing result is not a numeric value to compare.
Status vocabularies vary. Some sources use “BLQ,” “BQL,” or “<LLOQ”; some encode censoring in a result string rather than a separate field. Harmonize these codes according to a documented mapping, retaining the original source code. Do not infer that a blank result means below the LLOQ, or that a number printed at a limit is a genuine measurement. These meanings must be established from the source specification.
Worked Example: Distinguishing a Conflict From a Placeholder
A hypothetical assay has limits 0.05–5.00 ng/mL. Its documentation states that results below the LLOQ are stored numerically as 0.05 ng/mL with status “BLQ”; the number is a boundary placeholder, not a quantified measurement. Consider three records:
Record 1: 0.05 ng/mL, “BLQ.” The numeric value equals the documented placeholder and the status identifies the result as below range. This is consistent with the encoding rule; it should not be treated as a measured value of 0.05 ng/mL.
Record 2: 0.20 ng/mL, “BLQ.” Since \(0.05 \leq 0.20 \leq 5.00\), this value is within range, and it is not the documented placeholder. The status and value conflict. Flag the record for source review rather than changing its status automatically.
Record 3: 6.20 ng/mL, “above ULOQ.” Since \(6.20 > 5.00\), the status agrees with the result and limits. This check does not establish how the observation should later be used; it only confirms the consistency of its assay metadata and recorded status.
A Reproducible Validation Sequence
Match each result to metadata using the documented identity fields, such as analyte, matrix, method, laboratory, and version.
Use the documented date basis and interval rule. Require exactly one applicable metadata record.
Convert compatible units using a documented factor, then check that both limits exist and LLOQ is less than ULOQ.
Apply the assay's boundary and placeholder conventions. Distinguish quantified, below-range, above-range, and missing records.
Keep the source result and status traceable, and record whether the check passed, failed, or could not be evaluated, with a reason.
A short justification for the sequence is that each later check depends on the earlier ones. A comparison with limits is valid only if the limits belong to the correct assay version, apply on the relevant date, and use comparable units. If the lookup is ambiguous or the units cannot be reconciled, the range comparison has no defensible interpretation.
Diagnostics, Alternatives, and Limitations
Useful diagnostics include counting results with no matching metadata, more than one matching metadata row, missing limits, invalid ranges, unconvertible units, and status-value conflicts. Review patterns by assay version and source file: a sudden cluster of conflicts can indicate a changed code mapping or an incorrectly assigned effective date rather than many independent laboratory errors. Keep distinct outcomes for “consistent,” “conflict,” and “not assessable”; a missing limit must not be reported as a successful check.
Two common storage approaches are to keep limits in a separate assay-metadata table and join them during validation, or to copy the applicable limits onto every result row. A separate table avoids repeated values and makes method changes explicit, but requires a reliable, date-aware join. Copying limits onto result rows makes each row self-contained for review, but creates duplication and risks inconsistent updates. Either approach can work if the selected limits, source, and effective interval remain traceable.
These checks cannot verify the laboratory's underlying calibration, establish that the assay was performed correctly, or resolve an ambiguous source convention. They also do not decide how below- or above-range observations should enter a later analysis. They establish only whether the recorded result, assay identity, applicable limits, units, and censoring status are mutually consistent under documented rules.
Common Mistakes
- Using one limit for every assay. Limits are assay-specific; match the result to the appropriate analyte, matrix, method, version, and other required identifiers.
- Comparing numbers before checking units. Convert compatible units first and retain the conversion factor in the validation record.
- Choosing a date rule by convenience. Use the date basis specified for the assay, not whichever date happens to give a plausible result.
- Treating a placeholder as a measurement. A value stored at the LLOQ may encode a censored result. Check the source convention before interpreting it as quantified.
- Silently repairing a conflict. Preserve the source fields and flag the discrepancy for review; an automated check cannot establish which field is wrong.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- A result is 0.040 mg/L. The applicable limits are 30 and 800 ng/mL. Convert the result and decide whether it is within range.
- A quantified result is 0.12 mg/L with limits 0.10–3.00 mg/L. Under inclusive boundary rules, is its status consistent? Explain.
- A source stores 0.02 ng/mL with status “BLQ,” and its documented rule says BLQ values are stored at the LLOQ of 0.02 ng/mL. Is this necessarily a conflict? Why?
- Two metadata rows for the same assay identity both match a result's analysis date. What should the validation report say?
- An assay's limit changes on 1 July. The documented date basis is the analysis date, and the analysis occurred on 1 July. Which version should be selected if intervals are start-inclusive and end-exclusive?
Solutions. (1) \(0.040\ \text{mg/L} \times 1{,}000 = 40\ \text{ng/mL}\). Since \(30 \leq 40 \leq 800\), the result is within range. (2) Yes. The value satisfies \(0.10 \leq 0.12 \leq 3.00\), so it is within the quantification range. (3) Not necessarily. The documented placeholder rule explains why the numeric field equals the LLOQ; “BLQ” indicates the result is censored, not quantified at that value. (4) Report an ambiguous metadata match, not a successful range check; resolve the overlap or identity problem before comparing the result. (5) Select the version beginning on 1 July, because the start is included and the preceding interval ends immediately before that date.