The 90-Day Reassessment Framework: When and How to Adjust

Why Ninety Days

The interval is not arbitrary. Several independent timescales converge on it.

Glycaemic markers. HbA1c reflects roughly two to three months of glucose exposure, weighted to the most recent weeks. Re-measuring sooner samples the same period twice.

Membrane composition. The omega-3 index takes three to four months to reflect a new intake, since it measures red cell membrane fatty acids.

Trial durations. Most supplement trials with functional endpoints ran eight to twelve weeks. That is the window in which effects appeared, which makes it the fair window for a personal test.

Training adaptation. Mitochondrial and cardiovascular adaptations are measurable within four to twelve weeks; strength adaptations similar.

Vitamin D. Serum concentrations plateau on a new dose within eight to twelve weeks.

Shorter cycles produce noise and churn. Longer cycles let stale protocols run and let context changes go unnoticed for too long. Quarterly is where the biology and the practicality meet.

Question One: What Changed in Context?

This comes first because a context change can invalidate an entire protocol regardless of how the markers look.

• New or changed medication. The single most important item. Any new prescription requires an interaction re-check across the whole stack, and several common supplements affect drug metabolism through cytochrome P450 inhibition.
• Season and location. Vitamin D requirements change with both. A dose set in August is frequently wrong in February, and a relocation changes it within weeks.
• Training load. A change in volume or intensity changes protein needs, creatine relevance, magnesium losses through sweat, and the antioxidant timing question.
• Sleep and work pattern. Shift changes, travel and a new schedule change the constraint set entirely.
• Body composition change. Affects vitamin D dosing, hormonal profile and metabolic markers.
• Illness, injury or procedure. Changes inflammatory markers, invalidates comparison, and makes some compounds inappropriate temporarily, particularly those with antiplatelet activity before surgery.
• New clinical finding. Reframes the whole protocol and moves parts of it into clinical territory.
• Age bracket transitions and, for women, menopausal status. Slow-moving and consequential.

Nothing on this list announces itself in how a person feels, which is why it is a checklist rather than an intuition.

Question Two: Did Each Intervention Hit Its Objective?

This question is answerable only if the objective was written down at the start, which is the discipline the whole framework depends on.

For each item in the protocol, three things should have been recorded when it was added: the objective, the observation window, and what would count as evidence to stop. The review compares the result against those.

Where the objective was a measurement, the comparison is straightforward, with two cautions: use the same laboratory, and check whether the change exceeds the reference change value for that analyte, since a twenty percent shift in a variable marker can be noise.

Where the objective was a subjective outcome, such as sleep quality or afternoon energy, the comparison is weaker and expectation effects are real. A simple recorded note from the start beats recollection, which reliably reconstructs itself to match the decision already made.

Where the objective was mechanistic, meaning the compound was taken on the basis of plausible mechanism with no measurable endpoint, the honest verdict at ninety days is that nothing has been learned. That is a legitimate position to hold knowingly, and it should be labelled as such rather than defended as a result.

Question Three: What Comes Out?

This is the question almost nobody asks, and it is the highest-value part of the review.

Removal candidates, in order:

1. Anything whose objective you cannot state. If the reason is not recoverable, the item has no defence.
2. Anything that failed its own test. A fair window at an adequate dose with no change on the defined measure is a result. The response is removal, not dose escalation.
3. Anything whose objective has been resolved. A corrected vitamin D concentration means a maintenance dose, not the repletion dose. A resolved inflammatory source may make an anti-inflammatory addition redundant.
4. Duplicates within a functional slot. Two NAD+ precursors, three magnesium products, overlapping B vitamins across a multivitamin and a separate B-complex.
5. Anything now interacting with a new medication. Non-negotiable.
6. Anything whose evidence base moved against it. This requires someone to be tracking the literature, which is the least likely of these to happen unaided.

A useful test: if you were assembling this protocol today from scratch, with what you now know, would this item be in it? Anything that fails that question is being carried by inertia.

Question Four: What Is the Single Next Change?

One change per cycle, not five. This is the constraint that makes the framework work.

With several simultaneous changes, attribution becomes impossible: at the next review there is no way to know which change produced which effect, and the protocol becomes an accumulation of unfalsifiable decisions. With one change per cycle, every quarter produces a piece of information that is actually yours.

Choosing the change follows the constraint rather than the appetite. The question is which Pillar is currently limiting, and the review order is fairly consistent: sleep first, including any sign of disordered breathing; then cardiorespiratory fitness and muscle mass; then visceral adiposity and glycaemic handling; then inflammatory load; then nutrient adequacy; then everything else.

Most people reading about longevity are operating several steps below their actual constraint, which is why the single most common productive outcome of a first structured review is not a new compound but a change to training or sleep.

The Annual Layer

Some things do not need quarterly attention and do need attention.

Annually: a full laboratory panel rather than the quarterly subset; body composition; a fitness measure conducted properly rather than estimated; blood pressure across several readings; and a review of the whole protocol against the current evidence rather than against its original justification.

Every one to two years: an epigenetic or pace-of-aging measure if using one, with the same provider and method, since shorter intervals mostly measure assay noise.

Once: lipoprotein(a), which is largely genetic and changes little across a lifetime.

Age-appropriate clinical screening on its own schedule, which follows chronological age regardless of what any biological age measure says.

Why Protocols Decay

The deeper reason this framework matters is that a protocol is a snapshot of a moving system, and it starts going out of date immediately.

Every Pillar's status changes: with age, season, training load, travel, stress, illness, medication and body composition. Because the systems are coupled, a change in one propagates into others. The constraint that a protocol was built to address may have moved to a different Pillar entirely, and nothing about the protocol will indicate that.

Meanwhile the evidence moves. Compounds gain or lose support, regulatory positions change by jurisdiction, and interaction data accumulates. A protocol assembled from the best available evidence three years ago is not assembled from the best available evidence today.

So the useful unit of longevity practice is not the protocol. It is the process that keeps a protocol aligned with a changing person and a changing evidence base. Ninety days is simply the interval at which that process is cheap enough to sustain and frequent enough to matter.