What a Baseline Blood Test Should Measure, and How to Make the Numbers Usable

A baseline blood test is a panel drawn under controlled conditions, repeated at least twice within a few months, and stored as structured data with units and assay identifiers attached. The point is not the reference range printed on the report. The point is your own set point and your own variation around it, because that is what tells you whether a value three years from now has moved. A single draw compared against a population interval is a weak instrument. Two or three draws compared against each other is a real measurement.

Most commercial “baseline” panels fail on the second half of that definition. They give you a PDF, a population range, and no record of which assay ran the sample. That is enough for a screening conversation with a physician and not enough to support inference over a decade.

Why repetition matters more than panel size

Every analyte has three sources of scatter: analytical imprecision from the instrument (CV_A), within-person biological variation around your set point (CV_I), and between-person variation (CV_G). Only the first two affect whether your follow-up value differs from your own baseline. Studies that track healthy individuals across repeated draws exist precisely to estimate these components and to set analytical quality goals from them ¹.

The working tool is the reference change value:

RCV = 2^0.5 * Z * sqrt(CV_A^2 + CV_I^2)
# Z = 1.96 for 95% two-sided
# => RCV ~= 2.77 * sqrt(CV_A^2 + CV_I^2)

Plug in approximate literature values and the practical consequences are obvious. HbA1c has a low within-person CV, in the low single digits, so a shift of roughly 6 to 8 percent of the prior value is already outside noise. High-sensitivity CRP has a within-person CV in the tens of percent, which means a doubling can be pure biology. Triglycerides and TSH sit in between and are frequently over-interpreted. Before you react to any delta, compute the RCV for that analyte and check that your delta exceeds it.

Repeating the draw also lets you estimate your own CV_I rather than borrowing a published one. Three baseline draws spaced four to six weeks apart, all under the same protocol, give you a usable personal standard deviation for the stable analytes. That is the single highest-value thing you can do at the start.

The panel we would order

We would build the baseline in four blocks rather than buying a bundled “50 biomarker” product.

Block one is the standard clinical core. It anchors everything else and any physician can read it. Order a CBC with differential, a comprehensive metabolic panel, and a full lipid panel. Add liver enzymes ALT and AST plus ALP and GGT, along with creatinine and cystatin C together.

Block two is the cardiometabolic layer that standard panels often omit. Apolipoprotein B is the particle count and a better exposure measure than LDL-C when the two disagree. Lipoprotein(a) is largely genetically determined and needs measuring once in molar units (nmol/L) rather than mass units. Add HbA1c and fasting glucose, plus fasting insulin and hs-CRP.

Block three covers the common deficiency and endocrine axes. Order ferritin with transferrin saturation and vitamin B12 with either methylmalonic acid or homocysteine. Add 25-hydroxyvitamin D and TSH with free T4. For men, add total testosterone with SHBG drawn before 10am.

Block four is the optional frontier. It includes NT-proBNP, high-sensitivity troponin, and urate. Add IGF-1 and, where available, a plasma phosphorylated-tau assay. The last of these is worth understanding before ordering. Blood-based amyloid and tau assays now predict cognitive trajectory well enough to be studied in large clinic cohorts ², and immunoassay work has shown that plasma amyloid ratios carry real diagnostic signal ³. Population studies have also tracked remotely collected blood biomarkers against longitudinal cognitive outcomes ⁴. These are prognostic research measures in an asymptomatic person, and interpreting one in isolation without a clinician is a mistake.

Preanalytical control is where most baselines are lost

If the draw conditions change between visits, your delta measures the conditions rather than you. Fix a protocol and write it down.

Draw at the same time of day. Cortisol, testosterone, and iron all have diurnal swings large enough to swamp real change. Fast 12 hours with water allowed. That stabilizes triglycerides, insulin, and glucose. Skip hard exercise for 48 hours before the draw. CK, AST, and hs-CRP all respond to muscle damage. Sit quietly for 15 minutes before venipuncture, because standing shifts plasma water and concentrates protein-bound analytes by several percent. Hemoglobin and hematocrit are sensitive enough to plasma volume shifts that they can be used to model plasma and blood volume directly ⁵.

Tube handling matters as much as timing. Glucose in a plain serum tube drops roughly 5 to 7 percent per hour before centrifugation unless the tube contains fluoride or citrate. Hemolysis inflates potassium, LDH, and AST. High-dose biotin supplements interfere with many streptavidin-based immunoassays, including some thyroid and troponin platforms, so stop them for several days beforehand and tell the lab.

Store the data so you can compute on it

Ask for the structured export. Do not settle for the PDF. Most large US labs can return HL7 FHIR Observation resources or at minimum a CSV with LOINC codes. A LOINC code is a standard identifier for a specific measurement on a specific specimen, and it is what lets you join results from two different labs without guessing.

The schema we keep per result is:

collected_at,analyte,loinc,value,unit,ref_low,ref_high,lab,platform,method,fasting_hours,notes
2026-03-04T07:45:00-05:00,Ferritin,2276-4,68,ng/mL,30,400,Quest,Roche cobas e801,ECLIA,12,

The platform and method columns are the ones people drop and later regret. Ferritin, vitamin D, testosterone, and free T4 are not harmonized across manufacturers, and a change of platform can produce a 20 to 30 percent step change that looks like biology. When you switch labs, run one overlapping draw on both if you can afford it.

Guard against unit traps in code. The mg/dL versus mmol/L split affects cholesterol and triglycerides. The nmol/L versus mg/dL split affects Lp(a), where no exact conversion factor exists. Vitamin D is affected by the ng/mL versus nmol/L split. Store the unit as a string and convert at read time. Never convert at write time.

What a good baseline buys you later

Three concrete returns. First, deltas you can trust, because you know your own scatter. Second, a prior for interpreting any future abnormal result, which reduces the number of incidental findings that turn into cascades of imaging. Third, a usable pre-disease reference point if you ever need one. The clinical literature on treatment response leans heavily on pretreatment immune and inflammatory measurements, and in oncology baseline immune biomarkers are studied specifically as predictors of who responds to immunotherapy ⁶. Work in Alzheimer disease has followed the same logic, testing whether blood protein measurements taken before clinical diagnosis carry diagnostic information ⁷. Recent analyses extend this to metabolically at-risk groups, examining how blood biomarkers relate to cognition in people with diabetes and obesity ⁸.

None of that is a reason to self-interpret. A baseline is a measurement exercise. Any abnormal value, any RCV-exceeding change, and anything in the prognostic block belongs in front of a physician who can take a history and order confirmatory testing. We are describing how to measure well. We are not describing what any result means for you.

Questions people also ask

What is a baseline blood test? A set of blood measurements taken while you are well, under a fixed protocol, to serve as a comparison point for future results. Its value comes from being repeated and from being stored with enough metadata to compare across years.

Which tests should be in it? At minimum a CBC, a comprehensive metabolic panel, and a lipid panel with apolipoprotein B. Then HbA1c, fasting insulin, and hs-CRP. Then ferritin with transferrin saturation, TSH with free T4, and 25-hydroxyvitamin D. Lipoprotein(a) is worth measuring once.

Do I need to fast? Yes for triglycerides, fasting glucose, and insulin. Twelve hours of water-only fasting, repeated identically at each visit, removes a large source of between-visit noise.

How often should I repeat it? Two or three draws in the first three to four months to establish your set point and personal variation, then annually for the stable panel. Repeat sooner only for a specific question, and ideally with a clinician deciding what to recheck.

Can one abnormal result be trusted? Usually not. Compute the reference change value for that analyte and confirm with a second draw under the same protocol before treating the value as real.

Footnotes

1. Najib Aziz, Roger Detels, Joshua J. Quint, et al. Biological variation of immunological blood biomarkers in healthy individuals and quality goals for biomarker tests. BMC Immunology, 2019. https://doi.org/10.1186/s12865-019-0313-0 ↩

2. Vincent Planche, Vincent Bouteloup, Isabelle Pellegrin, et al. Validity and Performance of Blood Biomarkers for Alzheimer Disease to Predict Dementia Risk in a Large Clinic-Based Cohort. Neurology, 2023. https://doi.org/10.1212/wnl.0000000000201479 ↩

3. Andreas Nabers, Laura Perna, Julia Lange, et al. Amyloid blood biomarker detects Alzheimer’s disease. EMBO Molecular Medicine, 2018. https://doi.org/10.15252/emmm.201708763 ↩

4. Kumar B. Rajan, Neelum T. Aggarwal, Elizabeth A. McAninch, et al. Remote Blood Biomarkers of Longitudinal Cognitive Outcomes in a Population Study. Annals of Neurology, 2020. https://doi.org/10.1002/ana.25874 ↩

5. Louisa Margit Lobigs, Pierre‐Edouard Sottas, Pitre Collier Bourdon, et al. The use of biomarkers to describe plasma‐, red cell‐, and blood volume from a simple blood test. American Journal of Hematology, 2016. https://doi.org/10.1002/ajh.24577 ↩

6. Sacha Gnjatic, Vincenzo Bronte, Laura Rosa Brunet, et al. Identifying baseline immune-related biomarkers to predict clinical outcome of immunotherapy. Journal for ImmunoTherapy of Cancer, 2017. https://doi.org/10.1186/s40425-017-0243-4 ↩

7. James D. Doecke. Blood-Based Protein Biomarkers for Diagnosis of Alzheimer Disease. Archives of Neurology, 2012. https://doi.org/10.1001/archneurol.2012.1282 ↩

8. Michelle M. Mielke, Joni K. Evans, Rebecca H. Neiberg, et al. Alzheimer Disease Blood Biomarkers and Cognition Among Individuals With Diabetes and Overweight or Obesity. JAMA Network Open, 2025. https://doi.org/10.1001/jamanetworkopen.2024.58149 ↩