Reading a laboratory report: units, ranges and flags

• A laboratory report is a communication about one person's sample, not a diagnosis: it shows what was measured, the number, the unit, the laboratory's own reference interval, and any flag, so that a clinician can compare the result with the patient in front of them.
• The laboratory's own printed range always wins. Ranges change with method, instrument, age, sex, pregnancy and population, and ranges from a book or another laboratory are not reliable substitutes.
• A reference interval is built to include about 95% of healthy people. By design, about 1 in 20 healthy people falls outside it, and in a panel of 20 tests most healthy people will have at least one "abnormal" result.
• A flag (H or L) means "outside the laboratory's statistical range", not necessarily disease. The first thing to do is decide whether the result is real: haemolysis, lipaemia, an EDTA-contaminated sample, fist clenching during the draw and wrong patient are common culprits.
• Critical results are telephoned at once because they can indicate a life-threatening state. Trends, delta checks and the sample type and timing are usually more informative than a single isolated value.

By the end of this page you should be able to:

1. Read a report line by line and say what each field (patient and sample identifiers, test name, result, unit, reference interval, flag) is for.
2. Explain why SI and conventional units cannot be swapped carelessly, and convert the common analytes you meet on a report.
3. Explain how a reference interval is set, why about 5% of healthy people sit outside it, and why the laboratory's own interval is the one to use.
4. Define critical values, delta checks, and pre-analytical and analytical interference, and give original examples.
5. Interpret a short printed result sheet: step by step, with the numeric thresholds cited, ending in one clinical sentence and the next step.
6. Recognise the common report-writing traps (confusing units, reading another laboratory's range, ignoring sample condition) before they reach the patient.

This page is about the document, not the disease. Every topic in MedGrasp that uses a blood test assumes that you can first read the piece of paper (or the screen) that carries the number. A laboratory result passes through many hands and machines: a clinician decides that a measurement would help (the pre-preanalytical phase), someone checks identity, position, fasting and timing (pre-analytical), the analyser measures the analyte (analytical), the result is compared with a range and released (post-analytical), and someone acts on it (post-post-analytical).^[2] Errors can occur at every step, and up to 70% of laboratory errors occur in the pre-analytical phase, so the number on the screen is never the end of the reasoning.^[2]

Reading the report well is a skill with a small checklist. The Testing.com summary of report components is a good concrete picture of what clinicians and patients actually see on the page, and it matches the requirements that the Clinical Laboratory Improvement Amendments (CLIA) place on every report.^[4] Those required elements include patient name and identifier, name and address of the laboratory location, date the report was printed, date the result was generated, the name of the ordering clinician, specimen source and collection date and time, laboratory accession number, name of the test (often abbreviated), the result, any flag or abnormal/critical marker, units of measurement, the reference interval, an interpretation where appropriate, and any note on specimen condition or deviation from preparation instructions.^[4] A report that lacks the collection date and time, the specimen type or the accession number is harder to trust than one that carries them.

Good practice is to read a report in a fixed order: identify the patient and the date and time the sample was taken, note the specimen (whole blood, serum, plasma EDTA, urine, cerebrospinal fluid), read each line's number, unit and printed reference interval together so that the unit is never detached from the number, note any flag, check the specimen-condition line for haemolysis, lipaemia, icterus, clotting or insufficient volume, then decide whether the result is plausible, consistent with the patient and consistent with previous results for the same person.^[4][5] Time of collection matters for analytes with circadian rhythm: a midnight value for cortisol is expected to be low, so a "high" flag at midnight means something different from a "high" flag at 08:00.^[2]

"Reading a laboratory report" is not one measurement; it is the set of judgements that turn a stored quantity into information. Each line reports the amount of a specific chemical or cellular constituent, an analyte, in the submitted sample. The number is meaningful only in relation to a reference population, a unit, a sample type and the conditions under which it was collected.^[6] The report also carries information about how the sample was handled: a "specimen haemolysed" or "sample lipaemic" comment means the result may not represent the patient at all.^[4]

The flags and conventions on a report are worth memorising. A result inside the printed reference interval is often left blank or marked WNL (within normal limits). A result above the upper limit is often marked H and one below the lower limit L. An asterisk or similar mark is used for a critical (panic) value that has been telephoned to the ordering clinician.^[4] Negative and positive are used for qualitative tests where the answer is whether a substance or organism was detected at all.^[5] A cumulative report puts several tests or several days of results on one page so that trends and previous values can be compared, and using the same laboratory for serial monitoring reduces the apparent variation caused by different methods and instruments.^[1][5]

What the report cannot tell you is the diagnosis. A result outside the reference interval may or may not mean disease, and a result inside it does not guarantee health.^[5] Reference intervals are built from large groups of otherwise healthy people and overlap between health and disease, especially early disease.^[5] The clinicians' task is to put the number in its clinical context: history, examination, the pre-test probability of disease, the sample condition, other abnormal lines on the same sheet, and earlier results for the same patient.

A single line of a report is read as: result, unit, and the laboratory's own reference interval, side by side. Table 1 gives common analytes with the conventional unit range, the SI unit range and the source of each range used here. Ranges for sodium, glucose and creatinine are quoted from MedlinePlus encyclopedia pages for individual tests, which state the conversions explicitly.^[7][8][10] Ranges for any other analyte should be taken from the printed interval on the report in front of you, because methods and instruments differ between laboratories.^[1][6][9]

Conventional units are those still used in the United States and sometimes in the United Kingdom; SI units (Système International) are used in most other countries. The conversion is determined by the chemistry of the analyte: for monovalent ions such as sodium, potassium, chloride and bicarbonate, 1 mEq/L equals 1 mmol/L, so the number is unchanged. For substances measured by mass (glucose, creatinine, bilirubin, cholesterol), the conversion depends on the molecular weight. In practice, most laboratories print both, or state which convention they use in the column header.

The standard process for setting a reference interval is to measure the analyte in at least 120 otherwise healthy people from the relevant age, sex and population group, using a single method and instrument, then take the central 95% of the values and exclude the lowest 2.5% and highest 2.5%.^[1][6] This central-95% convention is not a subtle point: it means that by definition about 5% of healthy people (1 in 20) will have a result flagged as high or low.^[1] If 20 different tests are run on one healthy person, the chance of at least one abnormal result is 1 - (0.95)^20, which is 0.642, or about 64%.^[1] This is why a lone slightly abnormal result in an otherwise well person should not trigger a dozen new investigations. The Doles review also makes the caution explicit that only three analytes (cholesterol, creatinine and haemoglobin A1c) have undergone full international standardisation and harmonisation, so most methods genuinely give slightly different numbers; reference intervals are therefore labour-specific, not universal truths.^[1]

Validation is lighter than setting. A laboratory that adopts an interval from a manufacturer or from published data usually validates it with about 40 local samples rather than 120.^[6] Laboratories accredited by CAP, CLIA, COLA or the FDA are obliged to establish or validate their own intervals, which is why the same test in two different laboratories can carry different printed ranges.^[6] Elderly people, growing children, pregnant women, men and women have different intervals for many analytes; creatinine, haemoglobin, alkaline phosphatase, thyroid stimulating hormone and the blood counts are common examples where age and sex matter.^[1][6] A non-binary or transgender patient may need interpretation with both ranges in mind, since population data are still developing.^[1]

Practice in a global setting: write down the laboratory's own interval next to a result before deciding on an action. A result that looks abnormal against a memorised number can turn out to be inside the laboratory's printed interval, and vice versa. Repeating the test in an apparently well person using the same laboratory over time is often more informative than chasing one borderline value.^[1][5]

A flag points in a direction; it does not give the cause. Table 3 lists the broad categories that produce a high or low value, with the checks that separate them.^[2][5] Most "unexpectedly abnormal" reports are explained by one of the middle-row items rather than by new disease. Pre-analytical variables that shift results include fasting status, diet, time of day, posture, exercise, race, body mass, pregnancy, sex and medications; Table 1 of Zemlin's review lists these in three groups (patient, collection, handling) and they remain a useful mental checklist.^{[2] [3]}

Patterns on a report are named groups of lines that move together. Recognising them is usually faster and more reliable than reacting to one line:

• One slightly abnormal result among many normal ones in a well person is expected statistically; about 64% of healthy people have at least one abnormal value in a panel of 20 tests.^[1]
• Two or more analytes in the same panel drifting in the same direction (for example a rising creatinine with a rising urea, or a falling haemoglobin with a falling ferritin) is more likely to be biological than statistical.
• An isolated critical value, or one that conflicts with how the patient looks, should prompt a check of sample identity and condition before any management decision.
• A change large enough to fail the delta check (the comparison of the new result with a previous result for the same patient) should be questioned, even when both results are inside their intervals, because the patient cannot have changed that much for that analyte in that time.
• Haemolysis, lipaemia or icterus flagged in the sample-condition field casts doubt on electrolyte, enzyme and colour-based assays from the same tube.
• A specimen drawn downstream of an infusion, or one collected after a meal when fasting was requested, can produce a plausible but wrong set of numbers.^[3][9]
• Results obtained by different methods or different laboratories between visits are not the same time series and cannot be plotted together reliably.^[1][5]

Three panels. Every value is invented for teaching and labelled here as an illustrative sample, not real patient data. The interpretation after each sheet uses thresholds, ranges and causes from the cited sources above; thresholds are the ones stated by the cited pages (fasting glucose normal 70 to 99 mg/dL with prediabetes 100 to 125 mg/dL and diabetes 126 mg/dL or higher; sodium 135 to 145 mEq/L; creatinine men 0.7 to 1.3 mg/dL and women 0.5 to 0.95 mg/dL).^[7][8][10] Where a line's range is not quoted in the sources used here, it is printed on the sheet as supplied by the laboratory, because that is the only authoritative reference.^[5][6]

1. Sodium 140 is inside 135 to 145, fasting glucose 92 is inside 70 to 99, and creatinine 0.82 is inside 0.5 to 0.95, so no line is outside the laboratory's own interval.^[7][8][10]
2. No flag has been generated and a "normal" report with results inside the interval does not prove health, especially early in a disease, but it removes no surprises either.^[5]
3. Checking the sample-condition and collection-time fields adds confidence; a haemolysed or lipaemic sample would be noted there.^[4]

Conclusion: this report is unremarkable, and the next step is to interpret it alongside the clinical history rather than to repeat the tests.

1. Fasting glucose 118 is above 99 and below 126, placing it in the 100 to 125 mg/dL bracket that MedlinePlus classifies as impaired fasting glucose (prediabetes).^[7]
2. Sodium 141 is inside 135 to 145 and is not a problem in itself.^[10]
3. Creatinine 0.95 is inside the men's printed interval of 0.7 to 1.3 and does not suggest kidney impairment on this line.^[8]
4. A random glucose of 200 mg/dL or higher would also be abnormal under the same reference.^[7]

Conclusion: the pattern is prediabetes, and the next step is clinical review of risk factors with repeat fasting glucose or an oral glucose tolerance test and glycated haemoglobin as the clinician prefers.

1. Fasting glucose 246 is well above the 126 mg/dL (7 mmol/L) threshold that usually means diabetes.^[7]
2. Creatinine 2.4 is above the men's upper limit of 1.3 mg/dL (115 µmol/L), indicating reduced kidney function on this measurement.^[8]
3. Sodium 132 is below 135; MedlinePlus classifies a value below 135 mEq/L as hyponatraemia.^[10]
4. The three abnormalities together are a pattern rather than three isolated flags, and the sample condition line reassuringly does not report haemolysis or lipaemia.^[4]

Conclusion: this report points to diabetes with renal impairment and hyponatraemia, and the next step is urgent clinical review with HbA1c and a repeat renal panel rather than urgent unsupervised treatment.

• Treating a flag as a diagnosis. About 5% of healthy people fall outside the central 95% by construction. A lone mild abnormality in a well person is expected, not sinister.^[1]
• Reading another laboratory's range. Methods, instruments and populations differ, and only three analytes are fully harmonised internationally. Always use the printed interval on the report.^[1][6]
• Detaching the number from the unit. Convert (or let the report do it) before comparing. A creatinine of 2.4 mg/dL is far above the men's upper printed limit of 1.3 mg/dL (115 µmol/L); without the unit it is unreadable.^[8]
• Ignoring sample condition. A haemolysed or lipaemic sample invalidates several lines, and the comment line is the place to look. Up to 70% of laboratory errors occur in the pre-analytical phase, and 40 to 70% of rejected samples are rejected for haemolysis.^[2]
• Fist clenching and prolonged tourniquet. Both falsely raise potassium, and tourniquet use beyond one minute falsely raises albumin, calcium, glucose, proteins and several cell counts.^[2][3]
• Forgetting preparation. Fasting requirements (usually 8 to 12 hours, plain water only, no gum, smoking or exercise) exist because food and drink change circulating concentrations; a non-fasting sample reported as fasting creates a wrong diagnosis.^[9]
• Order-of-draw mix-up. K-EDTA carry-over from a lavender-topped tube can raise potassium and lower calcium. The CLSI-compatible order is blood culture, then light blue (citrate), red/yellow (serum), green/brown (heparin), lavender (EDTA), grey (glycolytic inhibitor).^[2][3]
• Drawing through an infusion or post-contrast. Dilution and iodinated or gadolinium contrast can interfere with creatinine, calcium, magnesium, troponin and iron. Where possible, collect at least two hours after contrast.^[2]
• Ignoring the delta. Result lower than the patient's previous trend by more than the accepted within-subject variation should prompt repeat or identity check.
• Overlooking the printed critical value. Laboratories must contact the requesting clinician immediately when a result indicates an immediately life-threatening condition or reaches the critical alert value; the report will show asterisks and the time of notification.^[4]
• Comparing across methods and laboratories. Even for harmonised analytes such as creatinine, the acceptable limits used in proficiency testing can differ by up to 35%; serial monitoring should stay with one laboratory.^[1]

• Read every result with its unit and the laboratory's own printed interval, and never swap in a memorised or foreign range.^[1][6]
• A reference interval contains about the central 95% of healthy people; about 5% of healthy people fall outside it by design, and about 64% of healthy people have at least one "abnormal" value in a panel of 20 tests.^[1]
• Intervals are set with at least 120 healthy locals for the relevant age, sex and method, and validated with about 40 samples when adopted; only cholesterol, creatinine and haemoglobin A1c are fully internationally harmonised.^[1][6]
• A flag means "outside the statistical range", not disease; the clinical context decides whether it matters.^[5]
• Strongly abnormal or critical results are telephoned to the ordering clinician at once, and the report records the time.^[4]
• Delta checks compare the new result with a previous one and prompt repeat or identity check when the change is implausible.
• Haemolysis (40 to 70% of rejected samples) and lipaemia (next most common endogenous interference) are the leading pre-analytical problems; both invalidate several lines.^[2]
• List pre-analytical variables systematically before blaming disease: fasting, time of day, posture, exercise, medications, tourniquet, draw site, tube, transport and temperature.^[2][3]
• Order of draw (blood culture, citrate, serum, heparin, EDTA, glycolytic inhibitor) prevents additive carry-over such as K-EDTA contamination.^[2][3]
• Fasting is usually 8 to 12 hours with plain water only; food and drink alter circulating constituents.^[9]
• A column of creatinine in mg/dL converts to µmol/L by multiplying by 88.4; glucose in mg/dL converts to mmol/L by multiplying by 0.0555; monovalent electrolytes are numerically identical.^[7][8]
• Serial trends are worth more than single values, and should be compared only when the laboratory and method are the same.^[1][5]
• The report is a tool for clinical reasoning, not a substitute for it; every flagged result finishes its journey in a consultation.^[5]