What Does “Not Detected” Mean in a Water Test Report?

ND is a result with a stated boundary

“Not detected” does not normally establish that the amount is exactly zero. First read the laboratory’s definition of ND, then the reporting limit and unit for that same substance. A result cannot be interpreted correctly by copying the two letters and discarding the rest of the row.

The complete evian annual report linked from its US website gives an unusually clear example. Its NSF laboratory notes define an ND result as a contaminant not detected at or above the reporting limit. That is the convention used in the examples below. Other laboratories may use different flags or distinguish detection, quantitation and reporting limits; their own legend controls.

The source is a historical manufacturer-published report, not a test carried out by this website. We inspected the identified rows, report headers and notes. We have not tested your bottle, home plumbing or private well, and these examples do not establish their condition.

Source: [1].

Keep the result and limit in the same row

In the report for sample S-0002187285, job A-00504667, the inorganic table on PDF page 4 shows arsenic as ND with a reporting limit of 0.001 mg/L. The report’s meaning is that arsenic was not detected at or above that reporting limit in the tested sample. Writing “arsenic = 0 mg/L” would add a claim that the report did not make.

In that same table, sodium is a numeric result of 7.1 mg/L with a reporting limit of 0.2 mg/L. The 0.2 is a reporting boundary, not a second sodium concentration and not the amount to use for a per-bottle calculation. The sodium reading guide shows that separate arithmetic task.

A numeric result close to a reporting limit also needs the laboratory’s interpretation. Do not independently reclassify an equality as ND, round a result across the boundary or infer measurement uncertainty from the number of decimal places. Preserve the reported form.

These are a few observations about fields in the complete source, not a reproduction of the laboratory report. Open the original PDF to see its entire table, footnotes and scope.

Source: [1].

ND, a blank standard and an absent test answer different questions

Scroll sideways inside the table to read every column.

A blank result is also different from a blank standard. Check the column heading. A missing field might require another page or clarification from the issuer. Do not assign the report’s explanation of a blank FDA standard to every empty cell.

The same PDF includes rows described as absent for particular microbiological tests. Keep that laboratory wording and its method rather than converting every non-numeric result into the same kind of ND concentration. “Absent,” N/A and a missing row are not interchangeable numeric values.

Source: [1].

Normalize units without turning ND into a measurement

For concentration arithmetic, 1 mg/L = 1,000 µg/L. Thus a reporting limit of 0.001 mg/L is 1 µg/L. The converted boundary is still a reporting limit; conversion does not produce a measured concentration.

Consider two invented results for the same named analyte: Report A says ND with a reporting limit of 0.001 mg/L, and Report B says ND with a reporting limit of 2 µg/L. After unit conversion, the reporting limits are 1 and 2 µg/L. These fictional observations show different reporting boundaries. They do not prove which sample had the lower actual concentration.

You cannot subtract the two limits and announce a 1 µg/L concentration improvement. You also cannot fill both ND values with zero and rank the waters by their equal “measured” concentrations. The required quantities were not reported.

A threshold comparison needs the same substance, concentration basis, unit and applicable rule. For example, a row expressed as nitrogen is not automatically comparable to an amount expressed as the whole nitrate ion. Preserve qualifiers such as “as N” and ask the report issuer when the basis does not match.

Source: [1].

Copy a complete interpretation record

REPORT AND SAMPLE
Original issuer URL / date checked:
Document title, issue date, page and job number:
Sample ID / product or sampling point:
ANALYTE ROW
Name, including any “as” qualifier:
Result exactly as written:
Reporting limit exactly as written / unit:
Legend defining ND or other flags:
Relevant standard column, jurisdiction and date:
INTERPRETATION
What this row actually establishes:
What is missing or cannot be inferred:
Question for the laboratory or responsible authority:

Record the sample identity before comparing results. This complete PDF includes more than one report and sample; reaching a later table does not mean the sample is unchanged. Copy its own header. The date on an annual-report cover is also not automatically the collection, analysis or authorization date of every sample.

If a reporting limit, legend or sample identity is missing, the useful outcome is a precise question to the laboratory. An incomplete interpretation should remain incomplete. For private wells, the well-testing guide explains how to obtain local testing and result follow-up.

Download the blank record as plain text. Keep missing information blank until the appropriate source supplies it.

Source: [1].

Three checks before using the result

Finally, a correctly read result does not decide a treatment response, clinical fluid allowance or current advisory. Follow the responsible local authority for the actual supply. The test-report and certification guide explains another separate limit: why a report’s PASS does not by itself authorize a certification mark.

When the question concerns an annual report statistic

An ND result is different from the range, average or compliance statistic reported by a water utility. Use the water-report reading questions to choose the appropriate guide: why a range does not identify a mean, highest TTHM samples versus the LRAA, or operational evaluation levels and their different calculation. These pages keep the actual report, monitoring conditions and current notices separate.