Design: Multicentre, parallel-group randomised controlled trial
Population: 732 women having early medical abortion, six New Zealand services
Primary outcome: Complete follow-up within 6 weeks
Effect: 94.5% vs 67.4%; RR 1.41 (95% CI 1.30–1.52)

A multicentre randomised trial in six New Zealand abortion services allocated 732 women having early medical abortion to self-assessment with a low-sensitivity urine pregnancy test or to comparative serum beta-hCG. The primary outcome was complete follow-up, meaning a test result discussed within six weeks.

Complete follow-up was 94.5% with self-assessment versus 67.4% with blood tests (RR 1.41, 95% CI 1.30–1.52). Successful abortion was recorded more often in the self-assessment group (89.0% vs 81.0%), though this likely reflects better ascertainment. Self-assessment led to more additional investigations (17.3% vs 11.1%), but complications, extra healthcare visits and contraception uptake did not differ. One ongoing pregnancy was found after the first trimester.

Serial serum hCG asks a woman to return for blood twice; many do not. A low-sensitivity urine test at home, with a planned phone review, kept far more women in follow-up without raising harm. Clinics still using routine blood tests could reasonably move to this model, with a clear route for anyone whose test stays positive.

Consider low-sensitivity urine self-testing with a scheduled phone review as the default after early medical abortion
Give clear written instructions on when to test and what a positive result means
Arrange a fast route to ultrasound or serum hCG when the self-test stays positive or symptoms persist
Discuss contraception at the first visit; follow-up method did not change uptake
Expect slightly more extra investigations with self-assessment and plan capacity for them

Why it matters

Loss to follow-up, not test accuracy, is the main weakness of blood-based checking after medical abortion.

The statistics, in plain English

A relative risk of 1.41 means complete follow-up was about 41% more likely with self-assessment; the confidence interval (1.30–1.52) is well clear of 1, so this is a firm result. Complication rates of 8.5% and 8.7% are effectively identical. A single ongoing pregnancy in 732 women is too few events to compare safety on that outcome between arms.