Editorial policy

What this policy covers

This page describes the working rules behind every calculator and guide on Baby Tools. It covers where our information comes from, how formulas are written and verified, how we treat units and assumptions, where we stop short of medicine, how often pages are reviewed, how AI is used, and what happens when we get something wrong.

The rules are deliberately modest. We are not a clinical publisher and we do not claim to be one. What we can control is whether our arithmetic is correct, whether our sources are public and checkable, and whether we correct ourselves openly when they are not.

Calculation principles

Four rules apply to every number this site displays.

The formula is public

Each tool shows the arithmetic it uses, in plain English and, where useful, as a short expression you could reproduce on paper. If two established methods exist, we name the one we use and say why we chose it, rather than blending them into a black box. A result you cannot reproduce is a result you cannot check.

Units are explicit

Weight is given in kilograms with pounds alongside, length in centimetres with inches alongside, and feeding volumes in millilitres with fluid ounces alongside. Conversions are stated rather than implied, and any rounding is described: we say how far a figure has been rounded, so you can tell the difference between 1,000 ml and "about a litre".

Assumptions are on the page

A due date calculated from a last period assumes a regular cycle and ovulation around day 14 of a 28-day cycle. An ovulation estimate assumes a luteal phase of roughly two weeks. Those assumptions are written next to the inputs, not buried in a separate document, because they are the usual reason a calculator disagrees with a clinician. Where an input lets you replace an assumption — a cycle-length field, for example — we explain what changes when you do.

Edge cases fail loudly

Invalid input produces a clear error, never a plausible-looking number. A date that does not exist, such as 30 February, is rejected. A last period dated in the future is rejected. A cycle length outside the range the method supports is rejected with the accepted range stated, rather than silently clamped to the nearest allowed value. If a result lands beyond the usual 40-week mark, we show how far past the date you are instead of hiding it. We would rather show you nothing than show you something wrong.

Where our information comes from

Sources are used in a fixed order of preference. The higher a source sits on this list, the more weight it carries when two sources disagree.

1. World Health Organization (WHO) Global guidance and reference standards, including growth references and definitions used across countries. who.int
2. Centers for Disease Control and Prevention (CDC) Public health guidance, surveillance definitions and growth charts used in the United States. cdc.gov
3. NHS Patient-facing guidance from the UK National Health Service, valued for its clear explanations of what happens at each stage of pregnancy and infancy. nhs.uk
4. ACOG and comparable specialty guidance Clinical guidance from the American College of Obstetricians and Gynecologists, and equivalent national obstetric or paediatric bodies. acog.org
5. Peer-reviewed research Individual studies indexed in PubMed, used mainly for ranges, definitions and the accuracy figures quoted on some pages. pubmed.ncbi.nlm.nih.gov
6. Other reputable secondary sources Medical textbooks, national statistics agencies and hospital patient-information pages, used only when the sources above say nothing specific and always labelled as secondary.

When sources conflict, we take the more conservative reading. In practice that means the wider range rather than the narrower one, the later milestone rather than the earlier one, and the wording that prompts you to ask a clinician rather than the wording that reassures you. A website has no ability to examine you, so the cautious version is the only defensible version.

Marketing pages, forum posts, influencer content and AI-generated summaries are never used as sources, however often they appear in search results.

Medical boundaries

Baby Tools calculates and explains. It does not practise medicine.

• No diagnosis. No page will tell you that you are pregnant, that a pregnancy is progressing normally, or that a symptom is harmless.
• No drug advice. We do not recommend, compare, dose or time medicines, supplements, herbs or treatments, including over-the-counter options.
• No test interpretation. We do not read hCG values, blood results, ultrasound reports, home test lines or growth measurements as a clinical finding.
• No false reassurance. If a value falls outside what our method can responsibly handle, the page says to contact a clinician instead of offering a cheerful interpretation.
• No replacement for care. Every tool page repeats that your midwife, obstetrician or paediatrician remains the decision-maker.

Some questions are simply not calculator questions. "Is this bleeding normal?", "should I be worried about this movement pattern?" and "which painkiller is safe?" all depend on examination and history. For those, the correct output is a sentence pointing you to a person, and we write that sentence rather than inventing a number to fill the space.

Review and updates

Every page is re-read against current guidance at least once a year. Topics where guidance moves faster — infant feeding volumes, growth references, screening schedules — are checked more often. When a major body such as the WHO, CDC, NHS or ACOG changes a recommendation that a page depends on, we update the page rather than waiting for its annual slot.

A review means three concrete things: reading the page beside the current source material, re-running a fixed set of test inputs through the calculator to confirm the outputs have not drifted, and checking that every outbound link still resolves to the document we cited. It does not mean clinical sign-off, and we do not describe it as one.

Every page carries a visible Last updated date. If that date is old, treat the page with corresponding caution — that is precisely why we show it rather than a vague "recently reviewed" badge.

How we use AI

AI tools help us draft explanatory text and write the code behind the calculators. Without them, a team this small could not maintain this many pages at this level of detail.

The limits are strict. AI is a drafting aid, not a source: it is never cited, never treated as evidence, and never allowed to supply a medical fact that a named guideline does not already state. A person checks every formula against the guidance before publication, verifies the arithmetic on sample inputs, and confirms every cited source by opening it. If an AI-assisted draft introduces an error, the error belongs to us — the tool does not absorb the responsibility.

We mention this because a site that hides its production method is asking for a kind of trust it has not earned. You are entitled to know that a human, not a model, signed off on the number in front of you.

Corrections and retractions

If you find a wrong number, a formula that does not match its cited source, or a claim stronger than the evidence, tell us through the contact link once the site launches. Include the page, your inputs and the result you expected; a reproducible report is resolved far faster than a general one.

Reports are handled in the order they arrive, with anything that could mislead someone about their pregnancy or their baby's health looked at first. We do not publish a reply-time promise we cannot keep. What we can promise is that a confirmed error is recorded in the update log on the affected page, in the same place where every other change is listed.

A page is withdrawn entirely — not merely annotated — when any of the following is true:

• The method it relies on has been withdrawn or contradicted by current guidance and no replacement method is settled.
• A defect affects a large share of results and cannot be corrected promptly.
• The sources behind a claim cannot be verified on re-check.

Withdrawal is noted in the update log so the gap is visible rather than mysterious.