Methods & consensus

Each calculator runs multiple published formulas and reports a consensus across them, rather than committing to one. A tdee call returns:

{
  results: [
    { method: "mifflin", value: 2759, unit: "kcal/day", detail: { … } },
    { method: "harris",  value: 2873.1, … },
  ],
  consensus: { mean: 2816.05, median: 2816.05, min: 2759, max: 2873.1, n: 2 },
  skipped: [
    { method: "katch", reason: "katch: requires body_fat or lean_mass" },
    { method: "cunningham", reason: "…" },
  ],
}

Methods whose inputs are missing appear in skipped (or throw in explicit mode). Provide body_fat or lean_mass to unlock the lean-mass-based methods.

Why a consensus instead of one number

Experts don’t agree on a single formula for most of these metrics — Mifflin-St Jeor and Harris-Benedict will hand you different BMRs from the same inputs, and both are “correct.” Rather than pick a winner and hide the disagreement, each tool runs every applicable formula and reports the spread. The mean/median is a reasonable working number; the min–max range is the honest uncertainty. When the methods cluster tightly you can trust the estimate more; when they spread out, that’s a signal to be cautious.

skipped is part of the same honesty: a method whose inputs are missing is listed with the reason, never silently dropped or guessed. Everything is deterministic — same input, same numbers, every time — so the consensus reflects formula disagreement, nothing else.

For how far to trust these numbers, see accuracy & limitations.