Sources

How we choose sources

The evidence hierarchy behind every range on this site, the labels we use, and how a range earns a stronger one.

The hierarchy

When sources disagree, we trust them in this order:

  1. Clinical guidelines and harmonized reference standards — the ADA Standards of Care, the ESC/EAS dyslipidaemia guidelines, Endocrine Society guidelines, and CDC-harmonized reference ranges.
  2. Large cohort and trial outcome studies — named, peer-reviewed, and cited next to the range they inform.
  3. Public reference libraries for test descriptions — MedlinePlus and the NIH Office of Dietary Supplements, linked from each test page under the publisher's own name.
  4. Founder judgment — used only to choose targets, and always labeled as ours. Judgment never outranks a guideline or a study.

The three labels

Clinical reference range
The interval a laboratory reports as normal, usually the central 95% of a reference population.
Outcome-associated range
A band a specific published study or guideline linked to a specific outcome. The citation travels with the number.
Founder-selected optimization target
A target we chose. Having targets is legitimate; calling them clinically proven would not be, so we never do.

Every range we use, labeled

Reviewed against the published literature on .

  • Clinical reference range
  • Outcome-associated range
  • Founder-selected optimization target

Used by the sample portal's illustrative score

  • Fasting glucose — 72–85 mg/dL
    Founder-selected optimization target

    Clinical reference: fasting glucose under 100 mg/dL is normal (ADA). In a 22-year cohort of nondiabetic men, cardiovascular mortality was higher above 85 mg/dL (Bjørnholt 1999); the largest cohort analysis found lowest cardiovascular risk at 85–99 mg/dL (Diabetes Care 2013). Evidence grade: partially supported.

  • HbA1c — 4.8–5.2%
    Founder-selected optimization target

    Clinical reference: under 5.7% is normal and 5.7–6.4% is prediabetes (ADA). In the ARIC cohort, risk was lowest around 5.0–5.5%, while values below 5.0% were associated with higher all-cause mortality (Selvin 2010) — our 4.8% floor is the weakest bound in the scoring set, and we flag it ourselves. Evidence grade: partially supported.

  • LDL cholesterol — 50–90 mg/dL
    Founder-selected optimization target

    Traditional reference: under 100 mg/dL is optimal (NCEP). The 2019 ESC/EAS guidelines set treatment goals below 70 mg/dL for high-risk and below 55 mg/dL for very-high-risk patients, and in the FOURIER trial benefit continued to a median of 30 mg/dL (Sabatine 2017) — so the ceiling is evidence-consistent and no evidence-based floor exists. Evidence grade: partially supported.

Benchmarks (never change your score)

  • Apolipoprotein B (ApoB) — at or below 80 mg/dL
    Outcome-associated range

    The secondary treatment goal the 2019 ESC/EAS dyslipidaemia guidelines set for high-risk patients; the very-high-risk goal is below 65 mg/dL (Mach 2019). Evidence grade: supported.

  • Total testosterone — 264–916 ng/dL
    Clinical reference range

    Clinical reference range for healthy young men: 264–916 ng/dL (Travison 2017, CDC-harmonized).

  • Free testosterone — 15–25 pg/mL
    Founder-selected optimization target

    This target is not interpretable until the assay method is named: reference intervals differ roughly tenfold between equilibrium dialysis (about 52–280 pg/mL) and direct analog immunoassay (about 9.3–26.5 pg/mL), and the Endocrine Society recommends dialysis or calculated free testosterone and advises against analog immunoassays (Bhasin 2018).

    Evidence grade: unsupported as stated.

  • Estradiol, sensitive (men) — 20–40 pg/mL
    Founder-selected optimization target

    Typical adult male reference: about 10–40 pg/mL. In men with chronic heart failure, mortality was lowest at 21.8–30.1 pg/mL and higher below about 13 and above about 37 pg/mL (Jankowska 2009); no optimal range has been established in healthy men. Evidence grade: partially supported.

  • Fasting insulin — 2.0–5.0 µIU/mL
    Founder-selected optimization target

    No guideline optimal range exists; the typical laboratory upper limit is about 25 µIU/mL. In a meta-analysis of 11 European cohorts, the highest fasting-insulin quartile carried 1.5–2.7× the cardiovascular mortality of lower quartiles (Hu 2004). Evidence grade: partially supported.

  • HOMA-IR — below 1.0
    Founder-selected optimization target

    A value of about 1.0 represents normal insulin sensitivity by construction in the original HOMA model (Matthews 1985). Published insulin-resistance cutoffs range from about 1.85 to 2.5 and higher depending on population and assay (Gayoso-Diz 2013); no guideline defines an optimal HOMA-IR. Evidence grade: partially supported.

  • Triglyceride:HDL ratio — at or below 1.0
    Founder-selected optimization target

    The published cut-point for flagging insulin resistance is 3.0 (McLaughlin 2003); risk rises continuously with the ratio (Gaziano 1997), and the ratio performs poorly as an insulin-resistance marker in African American populations (Sumner 2005). Evidence grade: partially supported.

  • TSH — 0.5–2.0 mIU/L
    Founder-selected optimization target

    Inside the standard reference range (about 0.4–4.5 mIU/L). Evidence within the normal range is contested: the HUNT cohort found higher coronary mortality at higher-normal TSH in women (Åsvold 2008), while a 14-cohort pooled analysis found no association (JAMA Internal Medicine 2015). Evidence grade: partially supported.

  • Free T3 — 3.2–4.2 pg/mL
    Founder-selected optimization target

    Inside the standard reference range (about 2.3–4.2 pg/mL). No outcome-based optimal free T3 range has been published for healthy adults; low free T3 predicts mortality mainly in people who are already ill. Evidence grade: unsupported.

The HbA1c benchmark above is the same range the score uses; both copies carry the same label and the same caution.

How a range advances

A range starts as a labeled target. It can be re-graded when a new guideline or a large cohort study speaks to it directly. It only earns the word "verified" after a licensed clinician reviews it and signs off — and until that happens, no range on this site is described as clinician-verified, including the three the score uses.

No clinician has yet reviewed these ranges. They are founder-selected targets and cited guideline goals, frozen as published here pending clinician sign-off. They are educational planning tools, not medical advice.

About the links on test pages

MedlinePlus, checked link by link

Most test pages link to MedlinePlus, the U.S. National Library of Medicine's patient reference, where a page for that test exists — every link was opened and checked on .

When a test links somewhere else — or nowhere

A few tests link instead to the source named on the link itself: the National Lipid Association consensus for ApoB, the NIH Office of Dietary Supplements for zinc, selenium and copper, and Frontiers in Nuclear Medicine for the NMR lipoprofile. Four tests — SHBG, cystatin C, fructosamine and DHT — carry no link, because we could not verify an authoritative page for them. We would rather show no link than a guessed one.

Your plan

Read full policy