Well, it's all because Charles Boklage did an amazing piece of research and statistical analysis in 1990. The results remain as our best statistic so far.
Preterm death of the human conceptus is common. A consistent biphasic pattern in the rate of loss from biochemical pregnancy detection to term suggests that most wastage occurs prior to clinical recognition. After simple adjustments for varying methods, existing data show that at least 73% of natural single conceptions have no real chance of surviving 6 weeks of gestation. Of the remainder, about 90% will survive to term. IVF conceptions do nearly as well as natural pregnancies after clinical recognition, but poorly before, despite selecting apparently normal embryos for transfer. Reasons may lie in the uterus more than the embryo itself. Multiple pregnancies may constitute more than 12% of all natural conceptions, of which number about 2% survive to term as twins and about 12% result in single births. In all of these situations, simple equations for exponential decay in a mixture of two populations can accurately describe the distribution of those deaths in time.So next time some stupid, ignorant sceptical person, pretending to know all about this subject, claims that the whole idea of womb twin survivors is baloney, give them this reference.