How to Read Blood Test Results and Catch What Matters

Start med kontekst, ikke de røde markeringene

Reference intervals usually capture the middle 95% of a comparison population, so about 1 in 20 results can fall outside range by chance alone. Patients trying to learn how to read blood test results often get misled because portals highlight the exception, not the pattern. If you are checking blood test results online, look for the lab notes before you look at the colored arrows.

Check the lab name, date, units, and fasting status before you interpret anything. Ferritin reported as 25 ng/mL and 25 µg/L is the same number, while vitamin D reported as 30 ng/mL equals 75 nmol/L — a conversion mix-up still fools smart people. Age, sex, altitude, pregnancy, and the specific analyzer can all shift what counts as expected.

Group results by body system instead of reading top to bottom. A lone ALT of 42 U/L may be noise, but ALT 42 U/L plus triglycerides 280 mg/dL, fasting glucose 108 mg/dL, and GGT 76 U/L points me toward metabolic liver stress much faster than the red arrow does.

As of April 24, 2026, this is also how we built Kantesti AI and why our background on About Us emphasizes clinician-led review. When I, Thomas Klein, MD, review an uploaded PDF, I look for timing, units, and clusters before I look at anything colored red.

Your 30-second first pass

First confirm your name, collection time, and whether the sample was fasting or marked hemolyzed. Then circle one panel at a time — CBC, chemistry, liver, lipids, iron or thyroid — before deciding whether any one number really deserves your attention.

Verifiser prøven, tidspunktet og skjulte preanalytiske fallgruver

Pre-analytic issues can change common markers before the sample ever reaches the analyzer. Hemolysis can falsely raise potassium by roughly 0.3 to 1.0 mmol/L and often nudges AST and LDH upward too, which is why I never panic over an isolated potassium of 5.6 mmol/L if the report mentions sample breakdown. EDTA contamination can do something even stranger — potassium looks high while calcium falls, a combination that should make you suspect the tube before the patient.

Timing matters more than most patients think. Morning cortisol is often highest around 6 to 8 a.m., testosterone is usually sampled before 10 a.m., and nonfasting triglycerides can run 20% to 30% higher than fasting values, so collection details belong next to the number, not in fine print. Patients who review fasting rules before blood work make fewer false assumptions.

Supplements can distort immunoassays. High-dose biotin, often 5 to 10 mg in hair and nail products, can push TSH falsely low and free T4 falsely high for 8 to 72 hours in susceptible assays; if that sounds familiar, read our note on biotin-related thyroid test errors.

The thing is, specimen marked hemolyzed, lipemic, or insufficient volume is not clerical trivia — it may be the clue that tells you to repeat the test before acting. I also see creatinine, albumin, and hematocrit look deceptively high after a long fast with poor hydration, especially in older adults.

Slik tolker du en CBC uten å miste mønsteret

Iron deficiency rarely starts with dramatic anemia. In practice I often catch it when hemoglobin is still 12.2 g/dL, MCV is 82 fL, and RDW creeps above 14.5%; that pattern usually sends me to ferritin long before the CBC turns obviously abnormal. Our page on oppfølging ved lavt hemoglobin walks through that sequence.

White cell clues sit in the differential, not just the total count. A WBC of 10.8 x10^9/L can be boring after stress or steroids, but neutrophils at 8.5 with lymphocytes at 0.7 tell a different story than eosinophils at 1.2 or circulating immature granulocytes. The CBC differential guide is the page I point patients to when they want to know which white cells actually changed.

Platelets above 450 x10^9/L are called thrombocytosis, but iron deficiency, recent infection, inflammation, splenectomy, and some cancers can all do it. Platelets below 150 x10^9/L deserve context as well; a stable count of 125 in a well person is a very different conversation from a fall from 240 to 125 in two months.

One easy miss: a normal hemoglobin does not rule out a developing problem. A 31-year-old with heavy periods, ferritin 11 ng/mL, platelets 430 x10^9/L, and hemoglobin 12.4 g/dL is often labeled normal in the portal even though the physiology is already telling on her.

Les kjemipanelet etter systemer: nyrer, salt og syre-base

Creatinine alone misses body-size context. A creatinine of 1.1 mg/dL may be ordinary in a muscular 28-year-old man and surprisingly high in a small 78-year-old woman, which is why renal panel interpretation should always include eGFR. An eGFR below 60 mL/min/1.73 m² for at least 3 months fits chronic kidney disease criteria, but one low value after dehydration does not.

Many labs now use the 2021 CKD-EPI race-free equation, and that change can shift eGFR by several points compared with older reports (Inker et al., 2021). When I see creatinine rise from 0.8 to 1.0 mg/dL while eGFR falls from 92 to 72, I care about the trend even if the portal still labels creatinine normal. Our article on tegn på høyt kreatinin covers the common non-kidney reasons too.

Sodium needs glucose beside it. Measured sodium usually falls by about 1.6 mEq/L for each 100 mg/dL glucose above 100, and some clinicians use 2.4 mEq/L when glucose is very high, so a sodium of 130 mEq/L with glucose 400 mg/dL may be much less alarming than it looks.

Kalium er svaret jeg har lest to ganger. Kalium over 6,0 mmol/L eller under 3,0 mmol/L kan forstyrre hjerterytmen, men falskt høyt kalium kan også forekomme ved hemolyse, knytting av hånden, svært høye trombocytter og uttalt leukocytose. Bikarbonat under 22 mmol/L med aniongap over omtrent 16 mEq/L fortjener også oppmerksomhet, fordi ketose, laktacidose, nyresvikt eller toksinpåvirkning flytter seg høyere på listen.

Lever, protein og bilirubin: mønsteret betyr vanligvis mer enn ALT alene

Leverenzymene fungerer best som et mønster. ALT og AST gjenspeiler hovedsakelig hepatocellulær irritasjon, mens ALP og GGT peker mer mot kolestatisk eller gallegangsmønster; vår veiledning for tolkning av leverprøver går dypere hvis panelet ditt inkluderer alle fire.

Vedvarende ALT eller AST over omtrent 2 ganger laboratoriets øvre referansegrense i mer enn 3 måneder bør vanligvis følges opp av lege, selv uten symptomer. En AST/ALT-ratio over 2 kan tyde på alkoholassosiert skade, men etter min erfaring blir dette klarere bare når GGT også er forhøyet og muskelskade er utelukket.

Bilirubin har sine egne regler. Totalt bilirubin er ofte 0,2 til 1,2 mg/dL hos voksne, og et isolert indirekte bilirubin på 1,5 til 3,0 mg/dL med normal ALT, AST, ALP, CBC og retikulocyttantall ender ofte som Gilbert syndrom, ikke leversvikt; vi forklarer dette mønsteret i vår bilirubinartikkel.

Albumin endrer seg langsommere enn enzymer. Albumin under 3,5 g/dL kan gjenspeile inflammasjon, tap av nyrefunksjon, leversyntetisk dysfunksjon, underernæring, fortynning fra IV-væsker, eller alt dette samtidig — og totalt kalsium kan se lavt ut når albumin er lavt, derfor er ionisert kalsium noen ganger et renere svar.

Når trening endrer AST mer enn leveren gjør

En 52 år gammel maratonløper med AST 89 U/L, ALT 31 U/L, normalt bilirubin og ømme quadriceps er ikke den samme pasienten som noen med AST 89 U/L, ALT 76 U/L, GGT 102 U/L og triglyserider 320 mg/dL. Hvis historien høres atletisk ut, pleier jeg å gjenta panelet etter 5 til 7 hviledager før jeg begynner å lete etter leversykdom.

Glukose og lipider: grenseverdier trenger kontekst, ikke gjetting

ADA standards still use fasting plasma glucose 126 mg/dL or higher, A1c 6.5% or higher, or a 2-hour OGTT glucose 200 mg/dL or higher to diagnose diabetes when confirmed or paired with classic symptoms (American Diabetes Association Professional Practice Committee, 2025). A1c between 5.7% and 6.4% fits prediabetes, but the A1c accuracy guide is essential if you have anemia, CKD, recent bleeding, or a hemoglobin variant.

Lipids also need risk context. LDL-C of 190 mg/dL or higher is treated as severe hypercholesterolemia in adults, and triglycerides of 500 mg/dL or higher raise pancreatitis risk; the lipid panel guide helps patients see how total cholesterol can distract from the more useful fractions. Non-HDL cholesterol is simply total cholesterol minus HDL, and it becomes especially helpful when triglycerides are elevated.

The 2018 AHA/ACC cholesterol guideline also treats apoB as a useful secondary marker when triglycerides exceed 200 mg/dL or risk feels disproportionate to LDL alone (Grundy et al., 2019). I see this pattern a lot in people with LDL 125 to 160 mg/dL, HDL 60 mg/dL, triglycerides 180 mg/dL, and a strong family history — numbers that look mediocre rather than urgent until you place them together.

Kantesti's neural network is particularly helpful when our AI blood test analyzer sees glucose, triglycerides, ALT, and uric acid leaning in the same insulin-resistance direction. One borderline number is easy to dismiss; five subtle ones moving together usually are not. That is also where our Family Health Risk feature becomes genuinely useful, especially in people who are still young enough to be falsely reassured by age.

Når 'normal' fortsatt skjuler jern-, B12- eller skjoldbruskkjertel-ledetråder

Ferritin is the iron-storage number most patients miss. Ferritin below 30 ng/mL often supports iron deficiency, and ferritin below 15 ng/mL makes it very likely, yet symptoms such as hair shedding, restless legs, and exertional fatigue can start before hemoglobin falls; that is why I still rely on low ferritin with normal hemoglobin more than a CBC alone.

Inflammation complicates ferritin. In inflammatory states, ferritin can sit between 30 and 100 ng/mL and still coexist with functional iron deficiency, especially if transferrin saturation is below 20%; this is one of those areas where context matters more than the neat lab cutoff.

Vitamin B12 is another trap. Levels below 200 pg/mL are usually deficient, 200 to 300 pg/mL is a gray zone, and patients can have numb feet, glossitis, memory fog, or mood change with a normal MCV. And oddly, B12 can read high in liver disease or some marrow disorders, so a high number is not always reassuring.

Thyroid panels are frequently oversimplified. A TSH of 4.8 mIU/L with low-normal free T4 means something different from a TSH of 4.8 with clearly normal free T4, positive TPO antibodies, pregnancy, or recent illness, and our thyroid panel explainer shows why free T4 and antibodies often change the conversation. Clinicians disagree on the exact upper cutoff, and some European labs use a lower upper reference around 4.0 mIU/L.

Bruk trender og ditt eget utgangspunkt før du får panikk

Trend reading is often where the real clue lives. A patient whose ferritin falls from 58 to 34 to 18 ng/mL over 14 months has a story, even if the first two reports looked fine in the portal; that is why I ask people to line up prior reports in date order or use year-over-year blood test history before they book the follow-up visit.

Every biomarker has noise, and the size of the noise varies. Triglycerides can vary 20% or more day to day, CRP can jump after a minor infection or dental procedure, and ALT often drifts 10% to 20% without implying new liver injury, so not every 3-point change deserves a new diagnosis.

What matters is whether the change beats your personal baseline and moves with related markers. A creatinine rise of 0.3 mg/dL, platelet drop of 50 x10^9/L, or new albumin fall of 0.4 g/dL gets my attention faster than a stable value that has lived mildly outside range for years; this is the logic behind personalized baselines. I, Thomas Klein, MD, worry more about a creatinine rise from 0.8 to 1.1 mg/dL than a bilirubin of 1.6 mg/dL that has sat unchanged for eight years.

Kantesti's trend analysis, built on clinician rules and described in our Medical Validation, compares assay-specific intervals, prior uploads, and multi-marker movement rather than red flags alone. Across 2M+ users in 127+ countries, our AI is most useful when it tells you not just what is abnormal, but what is new.

Hvilke blodprøver trenger oppfølging samme dag?

Urgent thresholds are not subtle. Potassium above 6.0 mmol/L, sodium below 125 mmol/L, glucose above 300 mg/dL with thirst or confusion, hemoglobin below 8 g/dL in many adults, platelets below 20 x10^9/L, and neutrophils below 0.5 x10^9/L usually deserve same-day medical contact. Our critical lab value guide explains why labs sometimes phone the clinician before the result even appears in your portal.

Symptoms can upgrade any result. A potassium of 5.8 mmol/L with palpitations, a troponin just above the assay cutoff with chest pressure, or a hemoglobin drop from 13.5 to 9.2 g/dL with black stools is more urgent than the raw number suggests; high potassium warning signs are a classic example.

Koagulasjonsprøver er en vanlig felle. INR, PT, aPTT, fibrinogen og D-dimer endrer betydning dramatisk hvis du bruker warfarin, heparin, apixaban, er gravid, nylig operert, eller håndterer kreftbehandling. En D-dimer under grensen for analysen hjelper mest når sannsynligheten før test er lav; en høy D-dimer er uspesifikk.

Tall er data; symptomer avgjør haster.

Hva du bør spørre legen om etter at du har lest rapporten

De beste oppfølgingsspørsmålene er enkle: Er dette nytt, hvilke andre markører støtter det, og hva ville endret behandlingen i dag? Hvis en lege ikke kan koble din avvikende ferritin, MCV, retikulocyttall, menstruasjonsforhold, kosthold og avføringshistorikk, har du fortsatt ikke et fullstendig svar. Personer som leter etter hvordan man tolker blodprøver, trenger vanligvis en skarpere liste med spørsmål mer enn en annen ordliste.

Ta med rapporten, de to foregående resultatene, en medisinliste, kosttilskudd og nøyaktig tidspunkt for prøvetaking. Når pasienter spør meg hvordan de skal forstå laboratorieresultater, vil jeg vite om de fastet, trente hardt, var syke, var dehydrert, brukte steroider, eller tok 5 mg biotin – slike detaljer endrer tolkningen mer enn de fleste forventer.

Brukt godt kan AI gjøre den kjedelige delen raskere og skjerpe de riktige spørsmålene. Vår artikkel om grenser for AI-tolkning av blodprøver forklarer blindsonene. For en første gjennomgang, AI-drevet tolkning av blodprøver på Kantesti kan hente ut en PDF eller et bilde på omtrent 60 sekunder, oversette enheter og oppsummere sannsynlige mønstre på tvers av 15,000+-biomarkører på 75+ språk ved hjelp av vår 2.78T-parameter Health AI. Vi bygde den under CE-merking, HIPAA, GDPR og ISO 27001 datakontroller, men den fungerer fortsatt best som beslutningsstøtte heller enn en diagnosemaskin.

Som Thomas Klein, MD, er jeg rett fram med pasientene her: ta med trenden, ikke bare den nyeste røde pilen. Hvis du vil teste arbeidsflyten før timen din, kan du prøve vår gratis demo. Hvis du vil se hvem som vurderer den medisinske logikken bak våre resultater, start med Medisinsk rådgivende styre.

De fleste pasienter går ikke glipp av ledetråder fordi de er uforsiktige; portaler fragmenterer historien. Kantesti ble bygget for å sette sammen den historien igjen ved hjelp av timing, trender, enheter og mønstre før du går inn til timen.

En kort liste å ta med til timen

Still fire konkrete spørsmål: Kan timing eller kosttilskudd forklare dette, hvilke relevante tester ville bekrefte eller avkrefte det, når bør det gjentas, og ved hvilken terskel bør jeg kontakte deg tidligere? Pasienter som stiller disse fire spørsmålene, går vanligvis derfra med datoer og utløsende punkter i stedet for vag beroligelse.

Frequently Asked Questions