Ovarian reserve and miscarriage risk outside assisted reproduction: a large cohort study

Miscarriage is the most common pregnancy complication and is associated with significant psychological morbidity. Previous studies, largely conducted in infertile patients and those undergoing assisted reproduction, have reported an association between low ovarian reserve and miscarriage. However, these findings may not generalise to women in the general population where evidence remains limited and inconsistent. This study investigates whether age-adjusted ovarian reserve is associated with a history of miscarriage among women accessing reproductive health testing services, independent of assisted reproduction.

This study leverages a large, structured digital reproductive health assessment capturing menstrual, symptom, lifestyle and medical history data to enable population-scale reproductive phenotyping. This was a retrospective cohort study of women who underwent diagnostic blood testing including AMH testing (September 2020-February 2025), outside of a fertility clinic setting. AMH was categorised into age-specific centiles (low, moderately low, within range, moderately high, high) and modelled as age-normalised z-scores. Pregnancy outcomes were grouped as live birth(s) only (LBO), miscarriage(s) only (MO), live birth(s) and miscarriage(s) (LBM), or never been pregnant (NBP). Binary logistic regression assessed associations between AMH range and history of miscarriage, adjusting for potential confounders. Miscarriage rate was calculated as miscarriages divided by total pregnancies and modelled against AMH z-scores.

Among 6,827 women reporting previous pregnancy, 4,141 (60.7 %) reported at least one miscarriage. AMH distribution differed across pregnancy history groups; however, associations observed in unadjusted analyses between moderately low or high AMH and miscarriage were attenuated after multivariable adjustment. Age-adjusted AMH z-scores demonstrated no clear relationship with miscarriage rate. Secondary analyses including women who had never been pregnant showed higher median AMH in this group.

In this large cohort, outside assisted reproduction settings, age-adjusted AMH was not independently associated with miscarriage history or miscarriage rate. These findings suggest ovarian reserve should not be used to retrospectively explain pregnancy loss or infer miscarriage risk in women outside ART settings.