Strong Diet

Biomarker Tracking

Summary

A short, well-chosen blood panel — ApoB, HbA1c with fasting glucose (and, if you can, fasting insulin), ferritin, vitamin D, hsCRP, and TSH — checked roughly yearly tells you almost everything a healthy person needs to know about their trajectory; the "full body" 100-marker panels sold by direct-to-consumer labs mostly generate false alarms, anxiety, and follow-on tests, while a handful of genuinely useful markers (ApoB, fasting insulin) are paradoxically under-ordered by routine care.

Why Strong

The entry is deliberately mixed-tier, assigned per marker, because lumping them would hide exactly the distinctions that make the panel useful.
• **The screening principle (small evidence-grounded panel > both no testing and full-panel) — Tier 1: backed by a large, consistent, independent overdiagnosis/DTC literature plus validated guideline diagnostics for the core markers. This is not a close call.
• ApoB, HbA1c, fasting glucose, ferritin (as screens) — Tier 1: large independent cohorts (apoB: ~590k+ participants), validated diagnostic thresholds (glycaemia), and RCT-level evidence for the treatable state (ferritin/iron-deficiency-without-anaemia).
• hsCRP in primary prevention — Tier 2: strong prognostic data and an RCT (JUPITER), but the pivotal trial carries an industry conflict, and the marker's confounding by acute inflammation limits clean use. NOT Tier 1 because of the conflict and confounding; NOT Tier 3 because the prognostic signal replicates independently (CANTOS).
• Fasting insulin / HOMA-IR as a routine screen — Tier 2-3:** mechanism and early-detection logic are strong (Tier-1-level physiology — insulin resistance genuinely precedes glucose change by 10-20 years), but assay standardisation and validated cut-offs are lacking and no guideline endorses it as a screen. NOT Tier 1 as a screen because of the standardisation gap; NOT dismissed because the early-warning value is real and aligns with recovery-before-crisis logic. (Read as a personal trend, its practical value is higher than its formal screening tier.)
• Vitamin D, TSH (as asymptomatic screens) — weak / Tier 2: USPSTF rates the evidence insufficient; over-testing and overtreatment are documented harms. Both retain Tier-1 diagnostic value when there's a symptom or a reason — the low tier applies to blanket screening of the well, not to targeted use.
• Lp(a) — Tier 1-2: strong causal/genetic risk evidence; once-only testing is well justified.

Practical takeaway

The starter panel (most healthy adults, ~yearly):
• ApoB (fasting not strictly required, but pair with a standard lipid panel)
• HbA1c + fasting glucose
• Ferritin (+ a CRP/hsCRP read to interpret it)
• hsCRP (only when you've been well for ~2 weeks — no recent infection, injury, or hard race)
• TSH — only if you have thyroid-suggestive symptoms (fatigue, cold intolerance, weight/hair/mood changes); not as a blanket screen
• 25-OH vitamin D — once to establish status (especially high-latitude / low-sun); then only seasonally if low or supplementing

Add if you can / if metabolically at-risk:
• Fasting insulin (to compute HOMA-IR) — the early metabolic warning, read as a trend against your own baseline
• Lp(a) — once, ever. Tick it off and never repeat it.

Frequency, honestly:
• For a healthy adult: roughly annual for the core metabolic/lipid markers is a sensible default. There is no strong trial evidence pinning the optimal interval for healthy people — most cadence advice (including this) is reasoned from how fast the markers actually change, not trial-derived. Flagged as such.
• If you're actively changing something (new training block, weight loss, a dietary overhaul, starting a lipid-lowering intervention), a re-test at 3-6 months is reasonable to see the effect — because you've created a real reason to expect movement.
• Do not re-test markers that don't move on your timescale. Lp(a) is genetic (once). Vitamin D moves over months, not weeks. HbA1c reflects 3 months by definition — re-testing it monthly measures nothing new.

What "working" looks like (the recovery read):
• ApoB trending down (if it was elevated and you're addressing it)
• HbA1c and fasting glucose stable in the healthy range; fasting insulin / HOMA-IR trending down is an early win that often precedes any glucose change
• Ferritin rising into a comfortable range (≥50 µg/L if you were fatigued and low) with energy improving
• hsCRP low (<1 mg/L is favourable; ≥2 mg/L is the risk-enhancer threshold) when measured well
• The honest non-event: most of your markers staying boringly stable year to year is the goal on a recovery trajectory. You are not trying to push numbers to extremes — you're confirming you're holding baseline.

How to read a result without panicking:
1. One abnormal value is a question, not an answer. Repeat it (after a few weeks, in a well/rested state) before you change anything or order more tests.
2. Read confounded markers in context — ferritin and hsCRP only mean something when you're not acutely inflamed.
3. Track against your own previous result, not just the reference range. Direction of travel > single threshold.
4. Resist the cascade. A borderline result does not automatically warrant the next, more invasive test. Ask: would this change what I do? If not, don't chase it.

What to skip: the "complete"/"full body" 50-100 marker panels marketed by DTC labs. The extra markers beyond the core set are overwhelmingly low-decision-value, high-false-positive, and exist to generate re-test subscriptions and the anxiety that drives them. (If a specific symptom points at a specific marker, test that — targeted, not blanket.)

Evidence detail

Why This Entry Exists

A Realised user wants to "get their bloods done" and is standing between two bad pieces of advice. On one side, a direct-to-consumer lab is advertising a "complete health panel" of 50-100 markers for a flat fee, with an app that turns every result into a red/amber/green dashboard and nudges them to re-test quarterly. On the other side, their GP runs a minimal panel, doesn't measure the two markers that would actually change their risk picture (ApoB and fasting insulin), and tells them everything is "normal" because each value sits inside a population reference range.

Both are failing the user, in opposite directions, and for the same underlying reason: neither is asking which markers carry decision-relevant information for a person on a recovery trajectory, and how often does that information actually change?

This entry exists to give the honest answer. It protects against the over-testing failure mode — the DTC "more data is always better" pitch, which is an industry that profits directly from your anxiety and from the cascade of follow-up tests a borderline result triggers. And it protects against the under-testing failure mode — the assumption that a standard NHS/insurance lipid panel and "you're in range" is the same as "you're fine," when the most predictive cardiovascular marker (ApoB) and the earliest metabolic warning (fasting insulin) are routinely left off the order.

What bad advice does this protect against?
• "Get the full panel — more markers, more insight." (Most of the extra markers have low decision value and high false-positive rates; the panel exists to sell re-tests.)
• "Everything's in range, so you're healthy." (Reference ranges are population distributions, not health targets; "normal LDL" can hide a high ApoB, and "normal glucose" can hide a decade of compensated insulin resistance.)
• "Track your biomarkers monthly to optimise." (Most markers don't move meaningfully on that timescale; frequent re-testing mostly measures assay noise and feeds anxiety.)
• "One bad result means something is wrong." (A single out-of-range value is often noise, an acute-phase effect, or regression-to-the-mean; the discipline is to repeat before you react.)

This is a Diet-pillar entry because the markers that matter most for a general user — the metabolic and lipid panel — are the readouts that diet, body composition, and activity move. It is the feedback loop for the whole platform's premise: you don't need to feel the change to verify you're returning toward baseline; a handful of numbers, read correctly, show it.

THE PANEL — what to test, why, and how to read it

This is the operational core of the entry. Each marker is rated for decision value (how likely a result is to change what you do) and given its honest evidence tier.

| Marker | What it tells you | Tier (as a screen) | Sensible frequency | The trap |
|---|---|---|---|---|
| ApoB | Number of atherogenic particles — the most accurate single lipid risk marker | 1 | 1-3 yearly; more often if treating | Often not ordered; "normal LDL-C" can mask high ApoB (discordance) |
| HbA1c | Average glucose over ~3 months | 1 | Yearly | Lags; normal until dysfunction is well advanced |
| Fasting glucose | Glucose at a moment | 1 | Yearly (with HbA1c) | Normal for years while insulin resistance builds |
| Fasting insulin / HOMA-IR | Earliest metabolic warning — how hard the pancreas is working | 2-3 (as routine screen) | 1-2 yearly if metabolically at-risk | No standardised cut-off; assay varies; not a guideline screen |
| Ferritin | Iron stores (+ fatigue/restless-legs context) | 1 | Yearly if symptomatic/menstruating/vegetarian | Acute-phase reactant — inflammation falsely raises it |
| Vitamin D (25-OH-D) | Vitamin D status | 2 | Once to establish; then seasonally if low/supplementing | Over-tested; assay-variable; screening healthy people is low-yield |
| hsCRP | Low-grade systemic inflammation / residual CV risk | 2 (primary prevention) | Yearly, but only when not acutely ill | Spikes with any infection/injury — useless if you have a cold |
| TSH | Thyroid function | 1 (if symptomatic) / weak (asymptomatic screen) | Only if symptomatic; not a routine asymptomatic screen | Borderline-high TSH normalises ~62% of the time on retest |
| Lp(a) | Genetic, lifelong CV risk particle | 1-2 | Once in a lifetime | Genetic — repeat testing is wasted money |

The rest of this entry is the detail behind that table.

Evidence

Read this in three groups: the markers with strong outcome evidence and high decision value (test these); the markers worth a targeted look but over-tested as blanket screens; and the screening principle itself.
Group A — high decision value, strong evidence (the core panel)

ApoB — the lipid marker that should be on every panel and usually isn't (Tier 1). Apolipoprotein B counts the actual number of atherogenic particles (each LDL, VLDL and Lp(a) particle carries exactly one apoB). Because cardiovascular risk tracks particle number more tightly than the cholesterol mass those particles carry, apoB outperforms LDL-C as a risk marker — and crucially, the two can be discordant: a person can have a reassuring LDL-C and a worrying apoB (small, cholesterol-poor, numerous particles), or vice versa.
• A 2024 systematic review of discordance studies (15 studies, 593,354 participants) found apoB superior to LDL-C as a risk marker in 9 of 9 direct comparisons (independent/academic synthesis; PubMed search to Sept 2024).
• A UK Biobank primary-prevention analysis (293,876 adults, free of CVD, ~11-year follow-up) found that when apoB was discordantly high relative to LDL-C, major adverse cardiovascular events rose (HR 1.11, 95% CI 1.06-1.15); when discordantly low, risk fell (HR 0.87, 95% CI 0.83-0.93). Elevated apoB independently predicted 20-year ASCVD risk regardless of non-HDL-C and Lp(a) (ATTICA cohort, 2002-2022). (Government/academic-funded cohorts; no commercial conflict on the marker itself — apoB assay is cheap and unpatented.)
• The honest caveat: apoB and LDL-C agree for most people most of the time. The marker earns its place not because it changes everyone's picture but because it changes the picture for the discordant minority — and you can't know in advance who that is without measuring it.

HbA1c + fasting glucose — the glycaemic floor (Tier 1). These are the validated, guideline-endorsed diagnostics for prediabetes and diabetes. HbA1c reflects ~3-month average glucose; fasting glucose is a single snapshot. They are reliable, standardised, and decision-relevant (they define the prediabetes/diabetes thresholds that trigger intervention). Their limitation is lateness: both can sit in the normal range for years while metabolic dysfunction develops upstream — which is the case for the next marker.

Ferritin — the highest-yield "tired all the time" marker (Tier 1, with one big caveat). Ferritin indexes iron stores and is the single most useful test for the extremely common, under-recognised state of iron deficiency without anaemia — low iron stores causing fatigue, poor exercise tolerance, and cognitive fog before haemoglobin drops enough to flag anaemia. A randomised trial in non-anaemic women with fatigue and ferritin <50 µg/L found 48% improvement in fatigue with iron vs 28.8% with placebo. (Independent/academic.) This is exactly Realised's territory: a cheap, treatable, mechanistically clear cause of the "low energy" so many users present with.
• The caveat that makes ferritin a trap if read naively: ferritin is an acute-phase reactant — inflammation, infection, recent intense exercise, or liver issues raise it independently of iron stores. So a "normal" or high ferritin can mask genuine iron deficiency in someone with any inflammation. The discipline: read ferritin alongside a marker of inflammation (CRP/hsCRP), and don't test it the week you have a cold or just after a hard race. (The WHO has revised its iron-deficiency ferritin cut-offs upward in inflammatory settings — from <15 µg/L toward <70 µg/L — precisely because of this.)
Group B — useful but targeted; over-tested as blanket screens

Fasting insulin / HOMA-IR — the earliest warning, and the most under-ordered (Tier 2-3 as a routine screen, but high decision value when positive). Insulin resistance develops 10-20 years before fasting glucose rises into the prediabetic range — because the pancreas compensates by pumping out more insulin to hold glucose normal. Fasting insulin (and the derived HOMA-IR = fasting glucose × fasting insulin / 22.5) can detect this compensation while HbA1c and glucose still read perfectly normal. Higher HOMA-IR is associated with greater subclinical atherosclerosis even after adjusting for traditional risk factors and HbA1c. (Independent/academic.)
• This is the marker that most embodies the under-testing failure: it is cheap, it sees the problem a decade early, and Realised's whole recovery before crisis premise argues for it — yet it is not recommended as a routine screen by major guidelines. Why the gap? Two honest reasons, not a conspiracy: (1) the assay is not well standardised and there is no universally validated cut-off (research uses ~2.0-2.5, but it's population-specific), and (2) there's no approved drug or guideline pathway that switches on at a high fasting insulin alone — so mainstream medicine "doesn't know what to do with it." Realised's position: it is genuinely informative as a direction-of-travel signal for someone working on metabolic health, read as a trend against their own baseline, not against a hard threshold. That is why it's in the panel but tiered as emerging as a screen — the mechanism and the early-detection logic are strong; the standardised-cutoff evidence is thin.

hsCRP — inflammation and residual cardiovascular risk (Tier 2 in primary prevention). High-sensitivity CRP is an independent cardiovascular risk marker with effect size comparable to blood pressure or cholesterol and good long-term reproducibility. The JUPITER trial showed that statins given to apparently healthy people with normal LDL but elevated hsCRP reduced cardiovascular events by 44% — establishing hsCRP as a real risk-stratifier, and ≥2 mg/L as a meaningful "risk enhancer." (JUPITER was funded by AstraZeneca, the statin's maker — an industry conflict that should be weighted; the marker's prognostic value, though, is replicated in independent cohorts including CANTOS.)
• The practical trap: hsCRP is wildly sensitive to any acute inflammation — a cold, a recent injury, a hard workout, a flare of anything will spike it. A single elevated reading in someone who's been unwell tells you nothing about chronic cardiovascular inflammation. It must be measured when the person is well, and a high value repeated to confirm before it means anything.

Vitamin D (25-OH-D) — worth knowing once, over-tested as a ritual (Tier 2). Knowing your vitamin D status once is reasonable, especially at higher latitudes or with little sun exposure (see light_exposure_vitamin_d). But the USPSTF concludes the evidence on screening asymptomatic adults is insufficient to determine the balance of benefits and harms, the assay is notoriously variable between labs, and 2024 USPSTF analysis found supplementation (with or without calcium) does not prevent falls or fractures in community-dwelling older adults. (Independent/government.) So: establish your level once, correct a genuine deficiency, re-check seasonally only if you were low or are supplementing — don't make it a quarterly ritual.

TSH — only if symptomatic; a poor blanket screen (Tier 1 diagnostically, weak as an asymptomatic screen). TSH is the right first test when thyroid symptoms are present (see thyroid_dysfunction). As a screen of asymptomatic people it is poor: the USPSTF finds insufficient evidence to screen asymptomatic adults, treatment of screen-detected subclinical disease has poor evidence of benefit, levothyroxine overtreatment is common (with real harms: atrial fibrillation, bone loss, cognitive effects), and — most tellingly — 62% of asymptomatic people with a single elevated TSH have a normal TSH on repeat. That last statistic is the entire over-testing problem in one number: test broadly, and most of your "abnormal" results are noise that resolves on retest, but not before triggering worry and a prescription. (Independent/government.)

Lp(a) — the once-in-a-lifetime test (Tier 1-2). Lipoprotein(a) is ~70-90% genetically determined and stable across life. High Lp(a) (≥50 mg/dL / ≥125 nmol/L) confers ~1.4-fold long-term heart-attack/stroke risk; ≥250 nmol/L roughly doubles it. The 2024 National Lipid Association update recommends measuring it at least once in every adult. Because lifestyle barely moves it, repeat testing is wasted. (Academic/professional-society guidance; note that targeted Lp(a)-lowering drugs are in trials, which gives industry an emerging interest in widespread testing — but the once-only logic and the causal-risk evidence predate that and stand independently.)
Group C — the screening principle (Tier 1)

The strongest evidence in this entry isn't about any single marker — it's about the strategy. A large, independent literature on overdiagnosis and direct-to-consumer testing converges on a clear finding: most DTC tests sold online have low clinical utility, target healthy consumers, and cause harm through false positives, overdiagnosis, anxiety, and diagnostic cascades (one borderline result triggering a chain of further tests, scans, and procedures, each with its own risk). A systematic review of DTC tests found 56% offered no pre/post-test consultation and 51% reported no analytical performance or lab accreditation. Overdiagnosis is defined precisely as detecting something that would never have caused symptoms or harm — meaning the affected person can only be harmed by the label. (Independent/academic; The Lancet 2024 characterised the DTC sector as "an industry built on fear.")

This is the counterweight to "more markers, more insight": past a well-chosen core panel, additional markers mostly add false-positive rate and anxiety, not decision value.

Mechanism

Why does a small panel beat a big one? The maths of testing, not biology.

1. Pre-test probability and the false-positive flood. Every test has a false-positive rate. When you run a marker in a population where the condition is rare (a healthy person, most markers), most "positive" results are false. Run 50 markers on a healthy person and — purely by the statistics of reference ranges, which are typically set so 5% of healthy people fall "out of range" by definition — you should expect two or three "abnormal" flags that mean nothing. The panel manufactures abnormality. A focused panel of high-prior, decision-relevant markers keeps the signal-to-noise ratio high.

2. Reference range ≠ health target. A lab "reference range" is the central 95% of a reference population — which, in a population where most people are metabolically drifting, includes a lot of suboptimal. "In range" means "not unusual," not "optimal" or even "healthy." This cuts both ways: it's why "normal" results can be falsely reassuring (your normal-range fasting glucose with a high-normal HbA1c and a high fasting insulin), and why being slightly "out of range" on a noisy marker is often meaningless.

3. Regression to the mean and biological variability. Any single measurement is one draw from a distribution that includes assay noise, diurnal variation, hydration, recent meals, recent exercise, and acute illness. An extreme reading is, statistically, likely to be closer to your true value on retest (regression to the mean). This is the mechanism behind "62% of elevated TSHs normalise on repeat" and behind why hsCRP and ferritin are so easily misread. The single most important testing discipline is: repeat an abnormal result before acting on it.

4. The acute-phase confounders. Ferritin and hsCRP (and to a lesser extent others) move with inflammation independently of what you're trying to measure. The mechanism is the acute-phase response: inflammatory cytokines (IL-6) drive hepatic synthesis of CRP and ferritin and raise hepcidin (which sequesters iron). So an infection or hard workout can simultaneously raise ferritin (hiding iron deficiency) and raise hsCRP (faking cardiovascular inflammation). Timing the draw to a well, rested state is not fussiness — it's the difference between signal and artefact.

5. The trajectory beats the threshold. The deepest reason a recovery platform cares about biomarkers is not the single value against a cut-off — it's the trend against your own prior result. Fasting insulin creeping up over two years (even within range) is more informative than any one reading. This is why a modest, repeated, consistent panel beats a huge, one-off panel: you're building a personal baseline to measure your own direction of travel.

Risks And Contraindications

The risks here are not from a needle — they're from misreading data and acting on noise.
• Health anxiety / over-monitoring. For some users, frequent testing and dashboard-watching becomes a source of chronic anxiety, not reassurance — every borderline value a fresh worry. If tracking is increasing distress rather than informing calm action, that is a signal to test less, not more. This is a genuine harm, evidenced in the overdiagnosis literature, and Realised treats it as one. Biomarker tracking is a recovery tool; if it becomes a compulsion, it has stopped serving the user.
• The diagnostic cascade. A single false-positive can trigger a chain of further tests, imaging, biopsies, and procedures — each with real physical risk (radiation, bleeding, infection) and cost — that would never have been needed. Restraint in chasing borderline results is protective, not lazy.
• Overtreatment. Screen-detected subclinical conditions (notably subclinical hypothyroidism) are frequently treated despite poor evidence of benefit, with real iatrogenic harms (levothyroxine overtreatment → atrial fibrillation, bone loss). Detecting something is not a mandate to treat it.
• False reassurance. The opposite failure: "all in range" read as "all is well." A normal-range result on a single marker does not rule out a problem the panel wasn't designed to catch, nor does it override symptoms. Numbers inform; they don't overrule how you actually function.
• Not a substitute for clinical care. Nothing here replaces a doctor. Genuinely abnormal, repeated results — and any acute or red-flag symptoms — need medical evaluation. Realised's role is to help a user choose a sensible panel and read it without panic, not to diagnose or treat.
• Fasting and timing errors. Lipid and glucose markers are affected by recent meals, alcohol, and exercise; ferritin and hsCRP by inflammation. A mistimed draw produces a misleading number. Follow the lab's prep instructions and don't test in the days after illness or a hard session.

Controversy

The core controversy: Should a healthy person test broadly ("more data is better, catch everything early") or minimally ("most testing in the well is low-yield and harmful")? This is a real, active dispute — and the honest answer rejects both poles.

Position A — "Test comprehensively and often" (direct-to-consumer labs, longevity-optimisation culture).
• Claim: large panels catch problems early, give people agency over their health, and "you can't manage what you don't measure."
• Real kernel: a few genuinely useful markers are under-ordered by mainstream care (ApoB, fasting insulin, Lp(a)), and proactive measurement of those is sound. Early detection of true metabolic drift is valuable.
• Where it over-reaches: extends "some markers are under-tested" to "all testing is good," sells 50-100 marker panels with low per-marker decision value and high false-positive rates, recommends frequent re-testing of markers that don't move, and rarely provides interpretation support (most DTC products offer no pre/post-test consult). The business model profits from both the test and the anxiety-driven re-test.

Position B — "Most screening of the well is low-yield and net-harmful" (USPSTF / overdiagnosis-prevention mainstream).
• Claim: for asymptomatic adults, the evidence for screening most individual markers (vitamin D, TSH, and many others) is insufficient or shows net harm via overdiagnosis, false positives, anxiety, and overtreatment.
• Real kernel: this is correct for blanket, indiscriminate screening, and the overdiagnosis evidence is strong.
• Where it under-reaches: guideline conservatism also leaves genuinely useful, cheap, early markers (ApoB, fasting insulin) off the standard order because there's no drug pathway that triggers on them — so "no screening recommendation" gets misread as "not worth knowing," when for a person actively working on their metabolic health the trend is decision-relevant.

The funding/bias dimension (and it cuts BOTH ways — this is the unusual part):
• On the over-testing side: DTC labs, "longevity" clinics, and biomarker-dashboard apps profit directly from selling large panels and recurring subscriptions. Their incentive is maximum markers, maximum frequency, maximum worry. The Lancet (2024) characterised the sector bluntly as "an industry built on fear." JUPITER, the trial that put hsCRP on the map, was funded by the maker of the statin it tested — an industry conflict to weight (though the marker's prognostic value replicates independently). Emerging Lp(a)-lowering drugs give industry a fresh interest in universal Lp(a) testing.
• On the under-testing side: the cheapest, most informative early markers (apoB assay, fasting insulin) are unpatented and unglamorous — there's no commercial champion funding their adoption, and no drug that switches on at a high fasting insulin, so guideline bodies have little pressure to recommend them. This is the classic "useful + unprofitable = under-investigated" signature, pointing toward under-use, not disproof.
• So the bias is genuinely two-sided: commerce inflates the volume of testing while neglecting the specific cheap markers that would actually help, because the money is in big panels and patented drugs, not in a £15 apoB and a £10 fasting insulin read as a trend.

Realised Position: A small, evidence-grounded core panel (ApoB, HbA1c + fasting glucose, ferritin, hsCRP, with fasting insulin and a one-time Lp(a) added where possible), tested roughly annually and read against your own baseline, is the honest middle. It captures the genuinely under-ordered markers that mainstream care misses, while refusing the DTC "full panel" that manufactures false alarms. The disciplines are: test the right few, time the draw well, repeat before reacting, track the trend, and don't chase the cascade. Testing is a recovery tool — if it's generating anxiety rather than calm, clear action, the dose is wrong.

Cross-Pillar Connections

• Diet / metabolic (blood_sugar_regulation, insulin_resistance_and_metabolic_dysfunction): the metabolic markers (HbA1c, fasting glucose, fasting insulin/HOMA-IR) are the direct readouts of these entries' subject matter. Fasting insulin is the early-warning surface for insulin resistance described there.
• Diet / cardiovascular (cardiovascular_health_management, cholesterol_misinformation_correction): ApoB, Lp(a), and hsCRP are the lipid/inflammatory risk markers; the cholesterol-misinformation entry owns the LDL-particle-vs-mass nuance that makes apoB superior, and this entry is its testing-practice complement.
• Diet / micronutrients (micronutrient_deficiency_screening, light_exposure_vitamin_d): ferritin and vitamin D are the micronutrient-status markers; the vitamin D entry owns the supplementation question, this entry owns when (not) to test.
• Mental / Physical (fatigue_cross_pillar_diagnostic): ferritin, vitamin D, TSH, and HbA1c are core to the fatigue work-up — iron-deficiency-without-anaemia is one of the highest-yield, most-missed causes of low energy.
• Mental (thyroid_dysfunction): TSH is the entry point; this entry's contribution is the discipline of symptom-triggered, repeated testing rather than blanket screening, and the 62%-normalise-on-retest caution.

What would change our mind

Falsifiability: explicit upgrade/downgrade criteria from source

We would broaden the recommended panel if:
• A marker currently outside the core set (e.g. a standardised insulin-resistance index, an emerging inflammatory or organ-specific marker) gained validated cut-offs and outcome evidence showing that acting on it in healthy people improves outcomes.
• Trial evidence emerged that more frequent testing of a specific marker in healthy people meaningfully improves outcomes (currently the cadence is reasoned, not trial-proven).

We would upgrade fasting insulin / HOMA-IR toward Tier 1 as a routine screen if:
• The assay were standardised across labs and a validated, population-appropriate cut-off were established, and a trial showed that intervening on high fasting insulin in normoglycaemic people improves hard outcomes.

We would narrow the panel / downgrade a marker if:
• A marker we recommend were shown, in well-powered studies, to produce more harm (via false positives and cascades) than benefit when used as a screen in healthy people.
• hsCRP's primary-prevention utility were shown to collapse once the industry-funded trials (JUPITER) are weighted out and independent replication thins.

What would NOT change our mind: DTC-lab marketing claiming a larger panel "finds more"; anecdotes of a rare condition caught by a broad panel (survivorship bias — it ignores the far larger number of false alarms and cascades the same broad testing causes).

Industry bias note

Structural incentives the evidence base may reflect

This topic carries a two-sided commercial distortion, which is rarer than the usual single-direction Realised pattern and is the main thing this section exists to name.
• Over-testing is sold. The direct-to-consumer lab and "longevity clinic" sector profits from large panels and recurring subscriptions; the incentive is maximum markers and maximum re-test frequency, and the product is often the anxiety as much as the data (no interpretation support, low per-marker decision value, false positives that drive the next purchase). The Lancet's "industry built on fear" framing is apt. Weight any "comprehensive panel" pitch against this.
• The pivotal inflammation trial is conflicted. hsCRP's clinical prominence rests substantially on JUPITER, funded by the statin's manufacturer. The marker's prognostic value does replicate independently — but the enthusiasm for screening-and-statinising on it carries a pharma fingerprint that should be weighted.
• Drug pathways shape which markers get recommended. Markers tend to enter guidelines when a treatment switches on at their threshold. This is why Lp(a) testing is rising now (Lp(a)-lowering drugs in trials) and why fasting insulin — cheap, early, informative — isn't recommended (nothing to prescribe at a high value). The recommendation landscape tracks treatability and patentability, not purely informativeness.
• The genuinely useful markers are the unprofitable ones. ApoB and fasting insulin are cheap, unpatented, and under-championed — the classic "useful but no revenue model" signature that points to under-use, not disproof. Realised's value-add is to put these cheap-and-informative markers into the panel while keeping the panel small.

Net: commerce inflates the volume of testing and neglects the specific cheap markers that would most help a recovery-focused user. The honest panel is small, includes the under-ordered cheap markers, and is read as a personal trend — not a dashboard to be maximised.

Sources (28)

Related entries

Blood Sugar Regulation and StabilisationStrongCardiovascular Health Management: Blood Pressure, Lipids, Clotting, and Heart HealthStrongCGMs for Non-Diabetics: A Biofeedback Toy, Not a DiagnosticEmergingCholesterol Misinformation CorrectionModerate
HIGH on which markers matter and how to read them; MODERATE on the optimal *frequency* for a healthy person (under-studied — most cadence advice is reasoned, not trial-derived).Last updated June 2026
Open in the Library: search, filter, every entry →

We set no cookies and run no ad trackers. We count visits with Cloudflare's cookieless, privacy-first analytics. The only thing stored on your device is which example you last viewed.