Why Do My Blood Test Results Vary Between Tests?
Biological variation, lab variation, and how to tell a real change from statistical noise — in plain English.
The Quick Answer
Every blood marker has two sources of variation. Biological variation is the natural fluctuation in your body day to day. Analytical variation is the imprecision of the lab measurement itself. Combine them and you get the reference change value (RCV) — the smallest difference between two results that is statistically real rather than random noise.
HbA1c has an RCV of about 6%. Ferritin is about 40%. ALT is about 50%. TSH is about 53%. So an HbA1c moving from 6.0 to 6.4 is probably real change; ferritin moving from 50 to 65 is probably just noise. Understanding RCV stops you over-reacting to small changes and helps you spot the meaningful ones. The same blood sample tested at two different labs can also legitimately give different numbers.
Two Kinds of Variation — Biology and the Lab
Biological variation is the natural day-to-day, hour-to-hour fluctuation in your body. Hormone levels shift through the day. Iron rises and falls with meals. Electrolytes shift with hydration. Inflammatory markers respond to any minor infection or injury. Even your posture matters — standing for 15 minutes before a draw shifts fluid into tissues and concentrates the blood by about 10%.
Analytical variation is the imprecision of the laboratory measurement. Even a perfectly calibrated, accredited Australian lab does not measure exactly the same number twice on the same blood sample. The coefficient of variation (CV%) of a method tells you how much repeat measurements scatter. For most modern automated tests the analytical CV% is small (2-7%) but it adds to biological variation.
The reference change value (RCV) combines both: it is the smallest difference between two results that should be considered a real change, with 95% confidence. RCV = 2.77 x sqrt(CV biological squared + CV analytical squared). When a change exceeds the RCV, it is statistically signal; when it is smaller, it is probably noise. This is the single most useful concept for interpreting repeat blood tests.
All the Sources of Variation — What Changes Your Results
Causes are grouped as biological (your body), pre-analytical (what happens between the draw and the lab analyser) or analytical (the lab method itself). Most patients only think about analytical, but biological and pre-analytical are the dominant sources in practice.
Time of day
Cortisol is highest at 6-8am, lowest at midnight. Testosterone is 30-50% higher in the morning. Iron and ferritin peak in morning. TSH is highest overnight and early morning.
Fasting state
Fasting glucose, triglycerides and insulin rise sharply after meals. Cholesterol is less affected. Most other tests are not affected by fasting state. Follow the pathology request form instructions.
Posture (standing vs supine)
Standing for 15 minutes before a blood draw increases protein-bound markers (albumin, calcium, total protein) and red cell counts by 5-10% due to fluid shifts. Most labs draw blood with you seated, which is a compromise.
Exercise in the prior 24-48 hours
Strenuous exercise raises creatine kinase (CK) up to 100x for days, lactate dehydrogenase (LDH), AST and ALT. Marathon runners can have CK levels 100x normal for a week. Avoid strenuous exercise 48 hours before testing.
Hydration state
Dehydration concentrates blood, mildly raising sodium, urea, creatinine, haematocrit and protein. Over-hydration dilutes them. Aim for normal hydration before testing — pale-straw urine the morning of the test.
Menstrual cycle
LH, FSH, oestrogen and progesterone change dramatically through the cycle. Most reproductive hormone tests need to be done on a specific cycle day (often day 2-5 for baseline). Pelvic ultrasound also varies through the cycle.
Recent illness or infection
Acute infection raises CRP, ESR and white cell count, and ferritin (an acute-phase reactant). It also reduces serum iron temporarily. Repeat tests 4-6 weeks after a significant illness for reliable interpretation.
Medication timing
Thyroxine taken just before blood draw spikes free T4. Statin timing changes lipid profile briefly. Iron supplements raise serum iron transiently. Ask your GP whether to take medications as usual or hold them before the test.
Alcohol in the prior 24-72 hours
Acute alcohol raises GGT for 24-48 hours and ALT/AST for a few days. Chronic alcohol raises MCV over weeks to months. For repeat liver function tests, avoid alcohol for 48-72 hours beforehand.
Sleep deprivation
Poor sleep raises morning cortisol and HbA1c progression, lowers testosterone, and can mildly elevate CRP and white cell counts. Try to maintain consistent sleep in the week before testing.
Lab method differences
Different analyser brands (Roche, Abbott, Siemens, Beckman) use different methods. Vitamin D, ferritin, thyroid antibodies and inflammatory markers differ most between brands. Use the same lab for comparisons.
Lab calibration drift
Australian labs are NATA-accredited and participate in external quality assurance through RCPA. Calibration drift is small but real, especially for low-volume tests. RCPA EQA programs catch most drift before it affects clinical care.
Sample transport and storage
Some markers degrade if sample sits too long before processing. Lactate, ammonia and certain hormones are most sensitive. Use the lab nearest to the collection centre when possible.
Haemolysis (broken red cells)
A traumatic blood draw can break red cells, releasing potassium, LDH and other intracellular contents. Labs flag haemolysed samples and may ask for a redraw. The most common cause of falsely high potassium.
Ethnicity-specific ranges
African-Australian patients tend to have higher CK and lower neutrophil counts than Anglo-Australian patients of the same age and sex. Asian populations have different glucose tolerance thresholds. Some lab ranges still do not reflect this.
Reference Change Values for Common Markers
CV biological is how much your own body varies day-to-day; CV analytical is the lab method imprecision; RCV is the threshold above which a change is statistically real. Values are approximate — exact figures depend on the lab method and patient context.
HbA1c
Most stable common marker. A HbA1c moving from 6.0 to 6.4 is probably real change; 6.0 to 6.2 is probably noise.
Sodium
Extremely stable. Even 2-3 mmol/L change is often meaningful, which is why hyponatraemia is taken seriously.
Creatinine
Day-to-day variation in creatinine of about 16% is normal. Trends across multiple tests matter more than single comparisons.
Total cholesterol
A cholesterol moving from 5.2 to 5.5 is within normal variation. A move from 5.2 to 6.5 (25%) is more likely real change.
eGFR
eGFR is calculated from creatinine, so it inherits the same variation. A drop of 10-15% over a single test could easily be noise. Look at the trend over 3-6 months.
Vitamin D (25-OH)
Very high variation. A vitamin D moving from 50 to 70 nmol/L (40%) is barely meaningful; a move from 50 to 90 nmol/L is real.
Ferritin
High biological variation, particularly because ferritin rises with any inflammation. Trend across 3 results is more reliable than a single change.
ALT
Liver enzymes vary a lot day to day. ALT moving from 30 to 45 (50%) is meaningful; 30 to 38 is probably noise. Recent exercise, alcohol or illness explain most variation.
TSH
TSH is logarithmic and highly variable. A TSH moving from 2.0 to 3.0 (50%) is barely meaningful; 2.0 to 4.5 is more likely real. Always interpret with free T4 if borderline.
Testosterone (total, male)
Highly diurnal and pulsatile. Always measure in the morning before 10am. Repeat measurements often differ by 30-40% even on the same week.
CRP (high-sensitivity)
Extreme biological variation because CRP responds to any inflammation. Trend over multiple tests is the only reliable interpretation; single results are nearly impossible to interpret without context.
Random glucose
Random glucose is affected by recent food. For consistent interpretation, use fasting glucose or HbA1c instead.
Situations Where Variation Is Exaggerated
In these contexts a single test can be misleading. Time the test outside these situations or repeat it once they have resolved for a more reliable picture.
Right after a hard workout
CK, LDH, AST, ALT and lactate can be markedly elevated for 24-72 hours after strenuous exercise. Marathon runners can have CK up to 100x normal for over a week. Avoid strenuous exercise for 48 hours before any blood test that includes muscle or liver markers.
After a stomach bug or vomiting
Dehydration concentrates blood and raises sodium, urea, creatinine and haematocrit. Sodium and potassium can shift. Repeat tests 1-2 weeks after recovery for reliable interpretation.
After a recent meal (when not fasting)
Glucose, insulin and triglycerides rise sharply. Iron rises if the meal contained red meat. Routine non-fasting tests are usually OK but glucose, lipids and iron studies are unreliable.
After an acute illness or infection
CRP, ESR, white cell count, ferritin and platelets rise. Iron drops. Liver enzymes can rise transiently. Repeat at least 4-6 weeks after recovery for stable values.
On hormonal contraception
Oral contraceptive pill raises SHBG, triglycerides, and some clotting factors. Free testosterone falls because SHBG rises. Interpretation of hormone panels needs to factor in the pill.
In pregnancy
Many markers shift substantially: ALP rises 2-3x (placental), creatinine falls, haemoglobin falls, triglycerides rise, thyroid demand increases, plasma volume expands. Reference ranges in pregnancy are different from non-pregnant.
After alcohol in the prior 24-72 hours
GGT rises within 24 hours and stays raised for 48-72 hours. ALT and AST rise for several days. MCV rises with chronic intake. For an accurate baseline, avoid alcohol for at least 72 hours before liver tests.
During acute psychological stress
Acute stress raises cortisol, glucose, white cell counts and blood pressure. Even fear of needles can raise these mildly. Repeat tests when you are calm if a result is borderline and you were anxious.
When Variation Crosses Into a Red Flag
Not all variation is benign. These patterns indicate the change has crossed from noise into signal and warrant a closer look or earlier follow-up.
Movement greater than the RCV in either direction
Any change exceeding the reference change value warrants attention. For example, an HbA1c jumping more than 6%, or an ALT jumping more than 50%, is statistically real and worth a closer look with your GP.
A trend moving in one direction across 3 or more tests
Even small changes consistently moving one direction over 3+ tests is more meaningful than a single big jump. eGFR slowly falling over a year is a kidney signal; HbA1c slowly rising is prediabetes drifting into diabetes.
Sudden large change that does not match how you feel
A large, unexpected change with no symptoms or lifestyle explanation might be lab error or pre-analytical issue. Ask whether the test should be repeated before acting on it — an out-of-range result is sometimes a redraw issue.
Conflicting results between two labs
If a result was abnormal at one lab and normal at another within a few days, raise this with your GP. Lab method differences can be the cause, but so can a real biological shift. Sometimes a third test resolves it.
Symptoms that do not match the numbers
If you feel significantly unwell despite normal results, or feel well despite abnormal results, the lab is not the whole story. Symptoms always warrant attention; ask whether additional testing or a specialist opinion would help.
A previously borderline result now well outside the range
A marker that was at the edge of normal and is now clearly abnormal warrants follow-up testing within weeks rather than the usual 3-12 months. Borderline-to-abnormal drift often signals a developing condition.
How to Standardise Your Tests — Reducing Noise
These seven steps standardise your testing conditions so the result reflects real biological change rather than noise.
Always use the same pathology lab
Different labs use different analysers and methods. For trend tracking, the same lab is the single biggest noise-reducer. Most major Australian providers (Sonic, Healius, Australian Clinical Labs) consistently use the same chain of labs for any given clinic, but check with reception when you arrive.
Always test at the same time of day
Aim for the same morning window (e.g. 8-10am) for repeat tests. Cortisol, testosterone, iron, ferritin and TSH all have a diurnal rhythm. A 4pm vs 8am draw can show a 30-50% difference for some markers.
Match the fasting state
If the first test was fasting (10-12 hours), the recheck should be fasting too. If it was non-fasting, the recheck can also be non-fasting. Water is fine during a fast; coffee, tea, gum and most supplements break it.
Match medication timing
Take thyroxine after the blood draw, not before, so you measure trough rather than peak. Take statins as usual. Stop biotin (in some hair-and-nail supplements) for 48 hours before any immunoassay test. Note any new medications on the request form.
Avoid strenuous exercise for 48 hours beforehand
Hard cardio or weights workouts raise CK, LDH, AST and ALT for days. If you train hard, time your blood test for a rest day. Light walking is fine.
Repeat at biologically meaningful intervals
Most chronic-disease markers (HbA1c, lipids, kidney function, thyroid) should be repeated at 3, 6 or 12 months unless a clinical reason demands sooner. Rechecking too often catches noise rather than signal.
Look at the trend across 2-3 results
Two data points define a line; three confirm a trend. For meaningful change, look for movement in the same direction across consecutive tests rather than reacting to a single result. Tools like SmarterBlood graph the trend automatically.
Food, Drink and Lifestyle — What to Match for Reliable Comparisons
Water in the morning of the test
HydrationMild hydration without overdoing it. Aim for 1-2 glasses in the hours before the draw. Concentrated blood from dehydration falsely raises sodium, urea, creatinine, haematocrit and proteins.
Consistent breakfast on non-fasting tests
ReproducibilityIf you are not fasting, eat a similar breakfast before each repeat test. Different meals lead to different glucose, triglyceride and iron values that confound comparison.
No red meat the night before iron studies
IronA heavy red meat meal in the 12-24 hours before iron studies can raise serum iron transiently. For consistent iron studies, choose a light meal the night before.
No alcohol 72 hours before liver tests
Liver enzymesAlcohol raises GGT, ALT and AST for days. Even moderate intake the night before will be visible in the result. 72 hours alcohol-free gives the cleanest baseline.
No biotin supplements 48 hours before tests
Hormone assaysBiotin (vitamin B7) interferes with many immunoassays including thyroid hormones, troponin, vitamin D and reproductive hormones. Hair-and-nail supplements often contain high-dose biotin.
No new high-dose B12 supplements before B12 test
B12 measurementRecent B12 tablets or injections can spike serum B12 levels artificially high, even when stores are depleted. Discuss timing with your GP if you have been supplementing.
Avoid coffee right before the test if fasting
Glucose accuracyCoffee mildly raises blood glucose and is generally counted as breaking a fast. Black water is fine; black coffee is not, especially for glucose tolerance testing.
Normal balanced eating in the week before
Marker stabilityBig short-term diet changes (like a 3-day juice cleanse or sudden keto switch) can move HbA1c indicators, lipids and electrolytes. Stick with your normal pattern in the week leading up to a recheck.
Related Reading
See Signal, Not Noise
Upload every blood test you have to SmarterBlood and the AI applies reference change value thresholds when showing your trend — so a 5% shift in HbA1c shows up as real, and a 30% wobble in ferritin shows up as noise. Spend less time worrying about small changes and more on the ones that actually matter.
This page provides general educational information about biological and analytical variation in Australian pathology testing. CV%s and RCV values are approximate and depend on the lab method, patient context and clinical scenario. This is not a substitute for professional medical advice. Always discuss meaningful changes in your blood test results with your GP. SmarterBlood does not provide medical care.