Comparing hormone concentrations across pregnancy requires attention to gestational age. This prospective study evaluated a booking-visit measurement while explicitly accounting for that changing background. Primary study
The cohort finding
Researchers recruited 993 asymptomatic pregnant women attending routine antenatal booking at six weeks' gestation or later. Lower gestation-adjusted plasma kisspeptin was associated with later miscarriage. The association persisted after adjustment for selected maternal characteristics. In this cohort, the reported discrimination estimate was higher for kisspeptin than for hCG. Primary report
Our appraisal: adjustment is essential but not sufficient
A gestational-age-adjusted result is different from a raw concentration cutoff. Applying one unadjusted number across several weeks could confuse normal developmental variation with a risk signal. The method used to build the gestational reference therefore matters.
Statistical adjustment for recorded characteristics does not establish causation. A lower concentration might reflect an underlying pregnancy process rather than cause the outcome. Raising the biomarker would not necessarily alter that process.
Discrimination describes ranking: whether a marker tends to separate outcomes within the studied sample. Calibration asks whether predicted probabilities match observed risks. A high discrimination estimate alone does not establish reliable absolute risk estimates in a new clinic.
The asymptomatic booking population also defines the setting. Performance may differ in people presenting with pain or bleeding, at earlier gestations, or after assisted reproduction. A test should be evaluated where it is intended to be used.
This cohort provides an observational signal worth validating. It does not establish that kisspeptin administration prevents pregnancy loss, that one result determines viability, or that usual assessment can be replaced. Full review should inspect the analytic population, outcome ascertainment, gestational reference model, thresholds, and independent validation. The distinction between a potential marker and a proven intervention is fundamental.
Frequently asked questions
Were participants asymptomatic at booking?
Yes.
Was this a prospective cohort?
Yes.
Were concentrations adjusted for gestational age?
Yes.
Does adjustment establish causation?
No.
Is discrimination the same as calibration?
No.
Was kisspeptin administered to prevent miscarriage?
No.
Why might another population need validation?
Its baseline risks, timing, symptoms, and measurement conditions can differ.