Methodology

Every effect row is entered by hand from the published literature, with range, latency, evidence grade and source. No entry is generated or estimated at runtime. Where a figure is uncertain, a weak evidence grade sits next to it, and where no curated figure exists yet, the entry is marked as an unconfirmed placeholder.

The catalogue in figures, counted from the same database this page serves.

The evidence grade describes how well the effect on the lab value is established. It says nothing about whether your clinical outcome changes with it. That sits separately in the outcome column.

For each measure, one row counts per marker you entered. The contribution grows with how far your value sits from the target range, with the point estimate against the largest known effect in the same unit, and with the evidence grade. Rows without a number keep a neutral size. Markers with no established treatment direction count with weight 0 and stay visible.

Evidence-grade weighting exactly as it stands in the code: A counts 1.0, B 0.8, C 0.5, D 0.3. The marker's urgency tier (1, 0.6, 0.3, 0.05) is applied as well. That lets you recompute any ordering here.

Big levers and small dials share one list. The order follows the evidence, not the category. A poor trade, where the cost sits on a more important marker, is shown in its own section.

Weight reduction drops out of the ranking when height and weight yield a BMI below 18.5. Without those figures the lever stays visible.

Whether an edge is benefit or cost follows from the effect direction and the marker's desirable direction. Markers without an established treatment direction (HDL, hs-CRP, homocysteine, vitamin D, total cholesterol) stay informational — never green progress. A row with direction none is a finding, not an empty cell.

Hidden raisers and prescription context explain a high value. They are for orientation, not lifestyle advice in the ranking.

The order follows benefit minus cost. On the card the two sides stay separate, rather than collapsing into one combined number. A poor trade goes in its own section. Whether an added purine load matters to you depends on your uric acid, your kidney function and your history of gout.

The search compares curated foods in the same class using recorded nutrients per 100 g. Suggestions are not a diagnosis. Composition comes from the German Nutrient Database BLS 4.0; purines come from the free USDA/ODS-NIH purine database (which includes Kaneko 2014). Qualitative reasons such as live cultures or a preparation are editorial notes, not effect rows.

When the page loads, the server delivers the anonymous effect database. Your values stay in a variable in the browser and are combined with that database there. Without an account they never flow back into a request. With an account the browser stores the history as ciphertext; plaintext figures are not in the database. The values are also absent from the address bar.

There is no tracking and no cookie banner, because there is no reason for one. To see whether the entry point works, we count three events. All that is stored per event and day is one number: how often it happened.

No lab value, no marker, no IP address and no identifier goes into these counters. So that one visit does not count several times, the browser remembers the event for that tab. We cannot tell from this who you are or what your report says.

The recorded target ranges with their source. Laboratory-specific reference ranges may differ; the range printed on your own report always takes precedence.

When we change something substantial in the database or in the presentation, it is listed here. Same log as on the home page.