How Accurate Are Blood Tests?
Australian pathology is among the most accurate in the world. But errors happen — mostly before the sample reaches the lab. Here is what can go wrong, how often, and how to spot it.
The Quick Answer
Australian pathology is among the most accurate in the world. All AU labs are NATA-accredited (ISO 15189) and participate in the RCPA Quality Assurance Program. Analytical accuracy for routine markers is typically 95-99 percentwithin a few percent of the true value — HbA1c around 2 percent CV, sodium under 1 percent, ALT around 5 percent.
The biggest source of error is not the lab. Around 70 percent of laboratory errors are pre-analytical — happening before the sample reaches the instrument. Haemolysis, lipaemia, wrong tube, prolonged tourniquet, sample mix-up, biotin supplements interfering with thyroid assays. Knowing these sources helps you spot a likely artefact and helps your GP know when to repeat.
The Three Phases Where Error Can Enter
A blood test result is the end of a chain. Error can enter at any of three stages. Pre-analytical covers everything before the lab analyser: patient preparation, the actual draw, transport, and centrifugation. Analyticalcovers what happens on the instrument: calibration, reagent quality, interference, method differences. Post-analytical covers what happens after the number comes off the instrument: transcription, the reference range applied, and interpretation in clinical context.
The pre-analytical phase is responsible for around 70 percent of laboratory errors, despite being the part most easily controlled by the patient. Eating before a fasting test, intense exercise the day before, prolonged tourniquet, or even sitting versus lying for the draw can shift values. The analytical phase is the most accurate stage — modern instruments are remarkable.
The post-analytical phase often produces the most consequential errors — a correct number with the wrong reference range, or missing the medication that interferes with the assay. This is where good clinical context (your GP knowing your full history) is most valuable. SmarterBlood adds another layer: applying RCV thresholds and biological variability calculations so you do not react to lab noise.
Sources of Inaccuracy — By Phase
Each error source is labelled with the phase it belongs to. Pre-analytical errors are the most preventable.
Haemolysis (red cells broken at draw)
Falsely high potassium, LDH, AST, magnesium. Lab usually flags this. Sample needs redraw.
Lipaemia (high triglycerides)
Interferes with optical assays. Falsely high or low for many markers. Repeat fasted.
Icterus (high bilirubin)
Interferes with optical methods. Affects creatinine, cholesterol assays.
Prolonged tourniquet (over 1 minute)
Falsely raises potassium, lactate, and calcium. Often the cause of an unexpectedly high potassium.
Wrong posture (lying vs upright)
Lying blood draws give lower haemoglobin, albumin, calcium, cholesterol than upright. Standardise position.
Time of day variability
Cortisol, iron and testosterone vary across the day. Always measure at the same time of day for serial comparison.
Recent food or drink (non-fasting when required)
Affects glucose, triglycerides, sometimes potassium. Lipids in particular need 12-14 hours fasting.
Recent exercise
Falsely raises CK, AST, LDH, urea, creatinine. Avoid intense exercise 48 hours before testing.
Medication or supplement interference
Biotin causes massive false thyroid and troponin errors. Many drugs interfere. Tell your GP and lab what you take.
Sample mix-up at collection or lab
Right blood, wrong patient label. Detected by major unexpected change or unusual marker pattern. Always repeat extraordinary surprise results.
Instrument drift or calibration error
Subtle systematic offset across many samples. Detected by QC programs. Lab investigates and reissues affected reports.
Reagent lot variation
New reagent lot gives slightly different values until calibration adjusted. Brief period of inaccuracy.
Method or assay interference (biotin, monoclonal proteins)
Specific interferents can cause large errors. Biotin is the classic patient-side issue.
Method differences between labs
Ferritin, free T3, prolactin and sex hormone assays differ between manufacturers. Same blood, different lab = different number.
Wrong reference range applied
Adult range applied to a paediatric sample, or vice versa. Always check the report header matches your demographics.
Transcription / clerical error
Decimal point shift, wrong unit. With electronic systems, very rare but worth checking against clinical sense.
Units misread by clinician or patient
mmol/L vs mg/dL on glucose, cholesterol, urea. Particularly when comparing AU and US references.
Missing clinical context
Result interpreted without knowing the patient is on warfarin, biotin, or in acute illness. Always tell your GP about medications.
How Precise Are Common Markers?
Each marker has an analytical CV (how variable the assay is on the same sample), a biological CV (how variable it naturally is from day to day within one person), and an RCV — the change between two results that exceeds noise and is meaningful.
Sodium
Tight regulation, very precise assay. Small changes are real.
Potassium
Heavily affected by haemolysis. Repeat if surprising.
Glucose (fasting)
Strict fasting required. Variable between days.
HbA1c
Very stable, makes it the best long-term diabetes marker.
Total cholesterol
Affected by stress, illness, weight loss in preceding weeks.
LDL cholesterol
Calculated value (Friedewald), so error compounds. Direct LDL is more accurate at high triglycerides.
Triglycerides
High day-to-day variability. Fasting status matters greatly.
ALT
Need substantial change to be meaningful. Small fluctuations are noise.
Creatinine
Day-to-day stable, affected by recent exercise and protein intake.
Ferritin
High biological variability plus acute phase reactant. Trends matter more than single values.
TSH
High day-to-day variability and diurnal rhythm. Test at the same time of day.
Vitamin D (25-OH)
Big method differences between labs and seasons. Trend within the same lab is most useful.
Free T4
Method differences between labs are significant.
Testosterone
Diurnal peak at 8am. Always test fasted at the same time.
Troponin (high-sensitivity)
Sensitive to biotin interference. Tell ED if you take biotin.
CRP
Massive biological variability. Used for clinical decisions only when clearly elevated.
Signs Your Result Might Be a Lab Error
These patterns deserve a repeat sample before acting on the result.
Result wildly different from your clinical picture
You feel well, no symptoms, no medications changed, but a marker is suddenly severely abnormal. Treat as possible lab artefact until repeated.
Single anomaly with otherwise normal panel
One marker dramatically out of range while everything else is normal often reflects a pre-analytical issue (haemolysis, lipaemia, wrong tube).
Severe abnormality but you feel fine
A potassium of 6.5 should make you feel terrible. If you feel completely fine, the result is more likely artefact than reality. Repeat before treating.
Repeat shows substantially different value
When repeating an unexpected result, a value that has moved by far more than the RCV between two close samples suggests one of them is wrong.
Wildly different result between two labs same week
Different methods can cause real differences, but a 50 percent gap on a stable marker like creatinine or sodium suggests one lab made an error. Worth a third tiebreaker test.
Result inconsistent with established trend
A marker that has been stable for years suddenly jumping massively without a clinical explanation deserves a repeat before acting.
Critical value triggering symptoms that do not match magnitude
A reported troponin of 500 ng/L (severe heart attack range) in a person with no chest pain or breathlessness deserves immediate repeat.
New medication or supplement before the test
Many drugs and supplements (notably biotin) interfere with assays. Always tell your GP and lab what you take.
Red Flags — When to Insist on Repeat
Most clinicians will repeat any of these patterns before acting. If your GP does not offer a repeat, asking for one is reasonable.
Critical result with symptoms not matching the magnitude
A reported sodium of 115 or potassium of 6.8 should produce dramatic symptoms. If a patient feels well, the result is more likely artefact than reality. Repeat before treating.
Same lab repeating returns substantially different result
When two consecutive samples a day apart return values differing by more than the RCV with no clinical change, one of them is likely artefact. A third sample tiebreaks.
Biotin user with abnormal thyroid or troponin
Biotin at high dose causes massive false errors on assays using streptavidin-biotin chemistry. Always ask the patient. Stop biotin and retest.
Sudden change in stable marker without clinical explanation
A creatinine that has been 90 for years suddenly jumping to 200 without symptoms, medication changes or kidney symptoms deserves a repeat sample, ideally with a fresh first-morning urine for urea-creatinine ratio.
Pattern of abnormalities that does not fit any disease
Diseases follow recognisable patterns. Random combinations of out-of-range values that do not fit any clinical syndrome are more often artefact than disease. A second sample usually clarifies.
Lab adds a comment "result probably due to interference"
Take this seriously. Australian pathologists actively flag suspect results. The recommendation will usually be to repeat with a new sample or different method.
How to Handle a Suspected Lab Error
A structured approach when a result does not match your clinical picture.
Repeat any unexpected critical result
Before acting on a severely abnormal but clinically surprising result, request a repeat. This is standard medical practice, not a sign of poor pathology. Ideally with a new draw at the same lab.
Ask about haemolysis on the original sample
The lab routinely checks for haemolysis. If the original sample was haemolysed, the report should state this and the affected markers (potassium, AST, LDH, magnesium, phosphate) should be repeated.
Try a different lab if doubt persists
If the same lab keeps returning an inconsistent value, request the same panel at a different NATA-accredited lab. Method differences explain some inter-lab variation; a large gap on a stable marker is a tiebreaker.
Calculate Reference Change Value (RCV) against your prior result
For meaningful change, the difference must exceed the RCV for that marker. SmarterBlood does this automatically. Smaller changes are within normal biological and analytical noise.
Check for medication or supplement interference
Biotin (any supplement above 5000 mcg) causes major false thyroid and troponin errors. Lithium and certain antibiotics also interfere. Stop the interfering substance and retest 1-2 weeks later.
Match the reference range to your demographics
Adult ranges applied to a paediatric sample, or sex-mismatched ranges, can make a normal value look abnormal. Check the report header lists your correct age and sex.
Bring all results together for clinical interpretation
A single number out of context is hard to interpret. Your GP weighs the result against your symptoms, medications, prior results, and the rest of the panel. This is where lab error is most often identified.
Quality Oversight in Australia
All Australian pathology labs operate under NATA accreditation to ISO 15189 (medical laboratories — requirements for quality and competence). The Royal College of Pathologists of Australasia (RCPA) runs the Quality Assurance Program (QAP), an external proficiency testing scheme where labs receive blinded samples and are scored against peers.
Labs that fall outside acceptable accuracy investigate and correct. Persistent failures result in loss of accreditation. Australia is consistently ranked in the top tier of pathology quality globally. If you genuinely suspect a lab error, your options are: (1) repeat at the same lab; (2) repeat at a different NATA-accredited lab; (3) raise with your GP, who can request a quality review from the original lab.
Preparation that Reduces Pre-Analytical Error
Plain water (during fasting window)
HydrationDehydration falsely raises urea, creatinine, sodium, and concentrates other markers. Drink plenty of water in the 24 hours before testing (but no other drinks during the fast).
Last meal at least 12 hours before fasting test
For lipids and glucoseA fast of 12-14 hours gives reliable fasting glucose and lipid panel results. Less than 8 hours can substantially alter triglycerides.
Standard diet for 3 days before testing
StabilityA sudden change in protein or carb intake just before testing can shift creatinine, BUN, glucose and ketones. Stick to your normal diet.
No alcohol for 24 hours before liver tests
Liver enzymesEven moderate drinking the night before raises GGT and can transiently raise ALT. Skip the night-before drinks to get a true picture.
No biotin / B-complex high-dose for 1 week before thyroid / troponin tests
Assay accuracyBiotin doses above 5000 mcg dramatically interfere with thyroid and troponin tests. Hair, skin and nail supplements are the common culprit.
Light dinner the night before intense workout (no exercise 48h before)
Muscle markersIntense exercise raises CK, AST, LDH, urea and creatinine for 48 hours. Take it easy the day before testing.
Avoid coffee / caffeine immediately before cortisol test
HormonesCaffeine transiently raises cortisol. For an accurate cortisol measurement, no caffeine for 4 hours before the test.
Continue all prescribed medications unless specifically told otherwise
Disease managementStopping medication for a test usually creates more problems than it solves. Tell the lab what you take so they can interpret in context.
Related Reading
Don't React to Lab Noise
SmarterBlood applies biological variability thresholds and reference change values to every result so you only see meaningful changes — not lab noise. Stop worrying about every small fluctuation.
This page provides general educational information about blood test accuracy and sources of variability in Australian pathology. It is not a substitute for professional medical advice, diagnosis or treatment. If you believe a specific result is incorrect, discuss with your GP, who can request repeat testing and quality review from the lab. SmarterBlood does not provide medical care.