Twin Probability Evidence Report

2 · Factor map

How the illustrative estimate was produced

Model 2.0 keeps a 0.40% monozygotic component stable and applies the selected weights only to a 0.85% dizygotic component. It then uses a 10% guardrail. The weights are conservative modelling choices informed by population literature; they are not causal effects, clinical coefficients or guarantees.

0.40% + (0.85% × 0.95 × 2.00 × 1.10 × 1.05) = 2.3%

3 · Uncertainty

Where this model is uncertain

• Population associations do not determine an individual outcome.
• The multiplicative method simplifies interactions between age and other population factors.
• The model separates baseline monozygotic and adjusted dizygotic components, but does not predict either outcome clinically.
• Fertility-treatment effects are excluded; patients need protocol- and clinic-specific data.
• The 10% guardrail is a modelling safeguard, not a biological threshold.

4 · Evidence context

Population data and definitions

Population statistics are useful context but cannot be inserted directly into a personal estimate. The Human Multiple Births Database counts twin deliveries per 1,000 deliveries; the US CDC counts individual twin babies per 1,000 births; the ONS reports multiple maternities. The full product will keep the source, year and denominator beside each fact.

5 · Source notes

Why each source appears in this report

A reference is not presented as proof that a personal outcome will occur. Each note states what the source contributes to the educational explanation and where the model remains uncertain.

1. [1]

   Calibration and US population context

   The CDC 2024 release is used as recent population context, not as a personal baseline or a direct model coefficient.

   CDC NCHS — Twin Birth Rates: United States, 2024 ↗
2. [2]

   Maternal age

   Age-related multifollicular development informs a deliberately compressed weight on the dizygotic component.

   Beemsterboer et al., Human Reproduction, 2006 ↗
3. [3]

   Biological family history

   Familial aggregation and genetic associations are stronger for spontaneous dizygotic than monozygotic twinning.

   Mbarek et al., American Journal of Human Genetics, 2016 ↗
4. [4]

   Population context

   Country and population differences are descriptive context and cannot classify or determine an individual outcome.

   Smits & Monden, PLOS ONE, 2011 ↗
5. [5]

   Height and BMI

   Observational associations are small, potentially confounded and compressed in model v2.

   Reddy et al., Obstetrics & Gynecology, 2005 ↗
6. [6]

   Previous pregnancies

   Parity is represented with a small heuristic weight and must not be interpreted as a recommendation to time pregnancies.

   Basso et al., JAMA, 2004 ↗