How to think about lab tests
A lab test is a measurement with error, not a verdict. Three ideas keep you out of trouble.
Reference ranges are population statistics. A lab's "normal" range usually covers the middle 95% of healthy people, so 5% of healthy people fall outside it on any given test. Run 20 independent tests and, by simple arithmetic (1 minus 0.95 to the 20th power), there is about a 64% chance at least one is flagged with nothing wrong. Ranges also differ because each lab validates its own, so compare results with the range printed on that lab's report.
Normal is not the same as optimal. A range describes the people tested, not the level linked to the lowest disease risk. Risk of heart disease climbs with HbA1c well below the diabetes cutoff, and LDL is linked to plaque even at levels labeled normal. The reverse also happens: values just outside a range are often harmless.
One value is a snapshot; a trend is a story. Many markers swing day to day. Urine albumin, for example, varies by more than 20% between samples, so guidelines require two of three abnormal tests over three to six months to confirm it, and diabetes is diagnosed only after two abnormal results unless blood sugar is unmistakably high. Use the same lab, the same time of day, and the same preparation when you can, and watch the direction over years.
- 64% chance that at least 1 of 20 independent tests is flagged in a healthy person, if each covers 95% of healthy people (reference ranges)
- 1 in 5 people have high Lp(a), which a single test reveals for life (NLA 2024)
What to test and how often
The optimal plan is a core panel every year from your 20s, so you have a baseline and catch change within a year rather than five. While you adjust a medication or make a major change in diet, weight, or training, repeat the tests it affects about every 3 months (lipids 4 to 12 weeks after each change). Each core test is inexpensive, standardized, and leads to a decision. No trial has compared testing intervals, so the yearly schedule is expert practice Weak; the guideline minimum is shown so you know what insurance covers.
The yearly core panel
| Test | What it tells you | Optimal | Guideline minimum |
|---|---|---|---|
| Lipid panel with ApoB | Plaque-causing particles | Yearly; 4–12 weeks after changes (heart) | Lipids at 19, then at least every 5 years; ApoB selectively |
| HbA1c and fasting glucose | Average and current blood sugar | Yearly; every 3 months while changing something | From 35, at least every 3 years; yearly with prediabetes |
| Fasting insulin | Insulin resistance before glucose rises | Yearly, with the glucose | None |
| Comprehensive metabolic panel | Electrolytes, liver enzymes, creatinine and eGFR | Yearly | None for healthy adults |
| FIB-4 (calculated from the panel and blood count) | Chance of liver scarring | Yearly from 35 | Every 1 to 2 years with metabolic risk factors (AASLD) |
| Cystatin C | Kidney filtering, unaffected by muscle mass | Yearly, with creatinine | When creatinine may mislead (KDIGO) |
| Urine albumin-to-creatinine ratio | Early kidney and blood vessel damage | Yearly | Yearly with diabetes; routinely with hypertension |
| Uric acid | Gout, kidney stone, and metabolic risk | Yearly | No screening guideline |
| Complete blood count | Anemia, infection, blood disorders | Yearly | None |
| hs-CRP | Low-grade inflammation | Yearly, when you are well | Selectively for heart risk |
| Vitamin D (25-hydroxyvitamin D) | Whether you need a supplement | Yearly; aim for about 30–50 ng/mL (supplements) | No routine testing (Endocrine Society 2024) |
| Ferritin, with iron and transferrin saturation | Iron deficiency or overload | Yearly if you menstruate, are vegetarian, train for endurance, or have heavy bleeding; a baseline for everyone | With symptoms or risk |
| Vitamin B12 | Deficiency that damages nerves and blood | Yearly from 50, or earlier on metformin or acid reducers or a vegan diet | Periodically on metformin (ADA) |
Several of these catch problems a basic panel misses. Adding cystatin C to creatinine found more kidney disease in 90,750 people (13.7% vs 9.7% with an eGFR under 60) and predicted death better. Iron deficiency (ferritin under 25) affected 38.6% of US females aged 12 to 21, most without anemia. A 2025 ACC scientific statement urged broad hs-CRP screening alongside LDL. The urine albumin test is cheap, often skipped, and an optional input to the PREVENT risk calculator.
Less often. A thyroid test (TSH) at about 35 and at least every five years after, or yearly if you like, since it is cheap (hormones) Weak; guideline minimum: the USPSTF finds insufficient evidence to screen. A baseline PSA for men by 45, or 40 at higher risk, repeated on a schedule set by the result. Lp(a) once in adulthood, ideally with your first lipid panel (heart), and HIV, hepatitis C, and hepatitis B once (below).
Optional extras
These are reasonable to add when they would answer a question, and each has a catch:
- Homocysteine, which flags low B12 or folate and kidney disease. Lowering it with B vitamins did not prevent heart attacks.
- Omega-3 index and sex hormones, when the result would change a decision (advanced tests, hormones).
What not to test routinely
Optimal is not maximal. Cheap tests are worth repeating; reflexively treating borderline results is where harm starts. Treatments have failed in trials when given for mild abnormalities or to people who were not deficient: levothyroxine for mildly underactive thyroids in 737 older adults, and vitamin D for 25,871 adults not selected for low levels. Test, confirm, and treat only a real deficiency. Epigenetic clocks, proteomic panels, and 100-marker subscriptions add flags faster than answers (advanced and trendy tests), and testosterone belongs to people with symptoms (hormones).
Biomarkers and targets
Each table gives the optimal target first, then the guideline cutoff. Where no outcome trial supports a target, it says so. Units are US conventional.
Lipids
These are the core of heart prevention; the reasoning behind the targets is in heart and blood vessels.
| Marker | What it tells you | Optimal target | Guideline cutoff |
|---|---|---|---|
| ApoB | Particle count; best single lipid marker | Under about 60 for most adults; lower with plaque or disease (expert) | NLA thresholds 90, 70, 60 by risk; Europe 100, 80, 65 |
| LDL-C | Cholesterol in LDL particles | Under about 70; under 55 with plaque (expert) | Goals under 100, 70, or 55 by risk (ACC/AHA 2026) |
| Non-HDL cholesterol | All plaque-causing cholesterol; free on any panel | Under about 100 (your LDL goal plus 30) | Goals under 130, 100, or 85 by risk |
| Lp(a) | Inherited particle; once is enough | No target; if high, lower everything else | Under 75 nmol/L low, 125 or more high |
| Triglycerides | Fat-carrying particles; tracks insulin resistance and alcohol | Under 100, which the AHA called optimal; a lifestyle target, not a drug target | Fasting 150 or more abnormal; non-fasting 175 or more |
| HDL-C | "Good" cholesterol; a marker, not a lever | Not a target; very high levels carry higher mortality | Under 40 (men) or 50 (women) is low |
| Triglyceride to HDL ratio | Rough insulin-resistance screen | Well under 3.0; 3.0 or more identified insulin-resistant adults | None |
ApoB and Lp(a) are not affected by whether you have eaten, and the other lipids change only slightly after a normal meal. For the full picture on Lp(a), see Lp(a) and inherited risk.
Blood sugar and insulin
Glucose problems start years before diabetes, and the guideline cutoffs mark diagnosis, not the start of risk. For what to do about high values, see metabolic health.
| Marker | What it tells you | Optimal target | Guideline cutoff |
|---|---|---|---|
| HbA1c | Average blood sugar over about 3 months | Under about 5.5%; risk was lowest at 5.0–5.5%, and 5.5–6.0% had 23% more heart disease | Under 5.7% normal, 5.7–6.4% prediabetes, 6.5% or more diabetes (ADA) |
| Fasting glucose | Blood sugar after 8 hours without calories | 70s to mid-80s; over 85 carried 40% more cardiovascular death in one 22-year study of men (observational) | Under 100 normal, 100–125 prediabetes, 126 or more diabetes |
| Fasting insulin | How hard the pancreas works to hold glucose down | Lower is better; 109 pmol/L (about 16–18 µIU/mL, by lab) or more flagged insulin resistance in one study | No guideline; assays are not standardized |
| HOMA-IR | Insulin resistance, calculated as glucose (mg/dL) times insulin, divided by 405 (Matthews 1985) | Under about 2; cutoffs of 1.9 to 2.1 matched metabolic syndrome in a Spanish population | No guideline |
Fasting insulin and HOMA-IR are early signals of insulin resistance, often abnormal while glucose and HbA1c look fine, but there is no outcome-tested target. HbA1c can mislead with anemia, iron deficiency, blood loss, kidney failure, pregnancy, and some hemoglobin variants, so a mismatch with glucose is worth a second look. A continuous glucose monitor shows day-to-day patterns; see wearables and home monitoring.
Inflammation, liver, and kidneys
| Marker | What it tells you | Optimal target | Guideline cutoff |
|---|---|---|---|
| hs-CRP | Low-grade inflammation | Under 1 mg/L, tested when you are well | Under 1 low risk, 1–3 average, over 3 high; repeat if over 10 |
| ALT (liver enzyme) | Liver cell injury, often from fatty liver | Under 33 (men) or 25 (women), the top of a truly healthy range | Lab limits vary |
| FIB-4 (calculated) | Chance of liver scarring, from age, AST, ALT, and platelets | Under 1.3, calculated yearly from 35 | Under 1.3 low risk; over 2.67 high; every 1 to 2 years with metabolic risk (AASLD) |
| Creatinine, cystatin C, and eGFR | Kidney filtering | A stable eGFR from the combined equation, which is more accurate | Under 60 for 3 months or more means chronic kidney disease |
| Urine albumin-to-creatinine ratio | Leaky kidney filters, a sign of blood vessel damage | Well under 30; risk rises within the normal range | Under 30 normal, 30–299 moderately increased, 300 or more severe (mg/g) |
| Uric acid | Gout, kidney stone, and metabolic risk | Under about 5.5 mg/dL through lifestyle; above about 5.6 predicted higher cardiovascular mortality (observational). No drugs without gout: allopurinol did not prevent heart attacks | No screening guideline; gout treatment aims for under 6 mg/dL |
| Homocysteine | An amino acid raised by low B12 or folate or by kidney disease | Optional; a high value prompts B12 and folate checks | No guideline; B vitamins lowered it but not heart attacks in 15 trials, with a small effect on stroke |
For kidney health and fatty liver, see metabolic health.
Blood count, iron, thyroid, and vitamins
| Marker | When it matters | Key thresholds |
|---|---|---|
| Complete blood count | Yearly baseline; fatigue, bleeding, infections | Lab ranges; changes over time matter more than one value |
| Ferritin and iron | Yearly with menstruation, heavy bleeding, a vegetarian diet, blood donation, or endurance training | With anemia, ferritin under 45 ng/mL points to iron deficiency (AGA); inflammation raises ferritin. High ferritin with high transferrin saturation prompts testing for hemochromatosis |
| TSH | At about 35, then at least every 5 years or yearly; sooner with symptoms or a family history | Repeat an abnormal result before treating; guideline minimum: no routine screening (USPSTF); see hormones |
| Vitamin B12 | Yearly from 50, or earlier with metformin, long-term acid reducers, or a vegan diet | The ADA advises periodic testing on metformin; deficiency grows more common with age |
| Vitamin D | Yearly; correct a low level and retest a few months later | Aim for about 30–50 ng/mL Weak; the National Academies judged 20 enough for bone and warned that levels above 50 raise concern. Guideline minimum: the 2024 Endocrine Society guideline sets no target and advises against routine testing; see supplements |
The metformin warning is real. In the Diabetes Prevention Program, low or borderline-low B12 affected 19.1% of people after 5 years on metformin vs 9.5% on placebo, and each year of use raised the odds of deficiency by 13%. A low-normal B12 with symptoms can be checked further with methylmalonic acid.
Hormones, PSA, and infection screens
Sex hormones. Testosterone should be measured in the morning, fasting, and repeated before anyone diagnoses a deficiency, because levels swing through the day. Estradiol, FSH, and other hormones are situational tests; interpretation and treatment live in hormones.
PSA. The optimal plan in cancer is a baseline PSA by 45 (40 at higher risk), repeated on a schedule set by the result. Guideline minimum: the USPSTF calls screening an individual decision for men 55 to 69, and an update is in progress as of October 2026.
One-time infection screens. Several infections stay silent for years, damage the liver or immune system, and are now treatable or curable. Here the optimal plan is the broadest guideline:
| Infection | Who | How often |
|---|---|---|
| HIV | Everyone 15 to 65 (USPSTF); the CDC says 13 to 64 | At least once; yearly or more with ongoing risk |
| Hepatitis C | Adults 18 to 79 (USPSTF); the CDC says all adults | Once; periodically with ongoing risk |
| Hepatitis B | All adults 18 and older with a three-part test (CDC 2023); the USPSTF recommends screening people at increased risk | Once; periodically with ongoing risk |
The hepatitis B "triple panel" also shows whether you are immune or need the vaccine. Sexually transmitted infections and vaccines are covered in infections and vaccines.
Genetic tests
Your genes do not change, so a genetic test is a one-time decision. The optimal plan in cancer is a clinical-grade multigene panel for every adult, ideally by 30, covering BRCA, Lynch syndrome, and other hereditary cancer genes Moderate; a panel that also covers the familial hypercholesterolemia genes adds little cost, and a US model found population testing for all three cost-effective under 40. Guideline minimum: testing only with a personal or family history. For any test, ask who else might see the result.
APOE and dementia risk
The APOE gene comes in three common versions. APOE4 raises the risk of Alzheimer's disease, and two copies raise it so much that a 2024 study argued APOE4 homozygosity is a distinct genetic form of the disease: nearly all people with two copies had Alzheimer's biology by 65, and symptoms began at an average age of 65.1.
For most people serious about prevention, knowing your APOE genotype is worth it Weak, with genetic counseling and after you have arranged any long-term care, disability, or life insurance you want, because the federal genetic nondiscrimination law does not cover those policies. A carrier result is a reason to start earlier and hold tighter targets, it informs decisions about trials and the newer Alzheimer's drugs, and in the REVEAL trial, adults told their genotype had no more anxiety or distress over a year. The caveats: there is no APOE-specific treatment, and the advice for carriers is mostly the advice for everyone. Guideline minimum: no guideline recommends APOE testing without symptoms. See should you get tested.
Familial hypercholesterolemia testing
If your LDL is 190 mg/dL or higher, or close relatives had heart disease young, a genetic test for familial hypercholesterolemia (FH) is worth asking about. An expert panel recommends testing the LDLR, APOB, and PCSK9 genes in anyone with probable FH and then testing relatives. A positive result confirms a lifelong risk, supports earlier and stronger treatment, and finds affected relatives before their first heart attack. See family history and familial hypercholesterolemia.
Polygenic risk scores
A polygenic risk score adds up the small effects of thousands of common gene variants into a single estimate of inherited risk for heart disease, diabetes, atrial fibrillation, and other conditions. The American Heart Association's 2022 statement offers only provisional guidance. In one large study it cites, adding a score to clinical risk factors nudged prediction accuracy (the C statistic) from 0.84 to 0.85, and most scores were built in people of European ancestry, so they travel less well to others. An Lp(a) measurement or a calcium scan usually tells you more about your heart than a polygenic score. Weak
Consumer DNA tests and privacy
Consumer genetic tests read a sample of variants, not your whole genome, and can miss disease-causing mutations that a clinical test would find, so a "negative" consumer result does not rule out an inherited condition. Treat any health finding as a prompt for clinical confirmation.
Privacy is the bigger issue. 23andMe filed for bankruptcy in March 2025, and that summer a court approved its sale to TTAM Research Institute, a nonprofit led by its co-founder, for $305 million, despite a lawsuit by 27 states and the District of Columbia seeking to block sale of genetic data without consent. Your DNA data is an asset that can change hands. Before testing, read the company's data and research policies. If you already have an account, download your raw data, decide whether to opt out of research, and consider deleting your sample and data.
Advanced and trendy tests
Longevity clinics and direct-to-consumer companies sell many tests beyond the core panel. An honest grading:
| Test | What it measures | Worth it? |
|---|---|---|
| ApoB | Plaque-causing particle count | Yes, every year; inexpensive and standardized Strong |
| LDL particle number (NMR) | Particle count, a different way | Mostly redundant with ApoB, which is standardized and cheaper Moderate |
| Omega-3 index | EPA plus DHA in red blood cells | Low values (4% or less) were linked to more coronary death than 8% or more; observational and rarely changes decisions Weak |
| Epigenetic age tests (for example TruDiagnostic) | DNA methylation "clocks" such as DunedinPACE | Research tools; experts have no consensus on clinical validation, and no result tells you what to do differently Weak |
| Proteomic aging panels | Thousands of blood proteins | A proteomic clock predicted death and disease in research cohorts; not validated for individual decisions Weak |
| Full-body "100 to 160 test" subscriptions | Many markers at once | Convenient if you have no doctor ordering tests; expect false alarms and redundant markers Mixed |
| Multi-cancer blood tests | Tumor DNA in blood | Whether they save lives is still being studied; see cancer |
The useful core of a big panel is the same handful of tests in this chapter. Breadth adds incidental flags (a slightly high bilirubin, a borderline vitamin), and each one invites a repeat test, a specialist visit, or worry. For biological age in general, see the science of aging.
How to get tests
Through your doctor and insurance. Ask your primary care clinician for the tests you want and say why ("my father had a heart attack at 52, so I would like ApoB and Lp(a)"). Under the Affordable Care Act, most plans cover preventive services such as cholesterol, diabetes, HIV, and hepatitis C screening at no cost in network for eligible adults, though extra tests and yearly repeats beyond the guideline schedule may be billed. For how to ask, see building your care team.
Direct to consumer. You can order tests yourself and have blood drawn at a commercial lab, with a contracted clinician signing the order. Prices as of October 2026:
- Quest (questhealth.com): a cholesterol panel for about $59, an iron and ferritin panel for $59, and an 85-marker profile for $399, drawn at more than 2,000 sites.
- Labcorp OnDemand (ondemand.labcorp.com): panels such as a 65-marker Advanced Health Panel for $399, at more than 2,000 sites or with home kits.
Subscriptions. Membership services bundle a large panel, an app, and a clinician summary:
- Function Health (functionhealth.com): $365 a year for more than 160 tests, split over two draws a year at Quest, with add-ons such as imaging.
- Superpower (superpower.com): $199 a year for more than 100 biomarkers in one annual draw, with add-on tests sold separately.
These are useful if your doctor will not order the yearly core, ApoB, or Lp(a), and two draws a year suit someone actively changing treatment. Their weakness is interpretation: a dashboard of 160 colored dots is not a plan. Bring the results to a clinician who knows you.
Preparing for a blood draw
A preparation checklist
- Fasting. Not needed for cholesterol, ApoB, or Lp(a) (EAS/EFLM). Needed for fasting glucose and insulin: 8 hours with no calories. Plain water is fine, and drinking it normally makes the draw easier.
- Biotin. High-dose biotin, common in hair and nail supplements, distorts some lab tests. At 10 mg a day for a week, it caused falsely high or low results in 9 of 23 tests that use biotin chemistry, including thyroid tests. Stop it for several days before testing and tell the lab you take it.
- Exercise. An hour of weightlifting raised the liver enzymes AST and ALT and the muscle enzyme CK in healthy men for at least 7 days. Avoid unusually hard or new workouts in the days before a test, or tell your clinician about them.
- Illness. Colds, infections, and injuries raise hs-CRP and ferritin. Test when you have been well for a couple of weeks.
- Timing. Testosterone and some other hormones should be measured in the morning. For trends, keep the same lab and time of day.
- Medications. Take regular medicines unless told otherwise, and list supplements on the order form.
When a result is abnormal
Most flagged results are not emergencies, and many are not real. Work through them in order:
- Check the size of the miss. A value just outside the range means something different from one far outside it.
- Check the context. Did you fast, exercise hard, take biotin, or have a cold? Is the value consistent with your past results?
- Repeat it. Many diagnoses formally require confirmation, such as two abnormal tests for diabetes and two of three for urine albumin.
- Ask what would change. A good next step is one that leads to a decision: a medication, a referral, a lifestyle change, or a planned recheck.
- Act on the important ones. Persistently high ApoB, LDL, HbA1c, blood pressure, or urine albumin are reasons to treat, not just to keep testing.