It’s a popular teen pregnancy prevention tool used in high schools in at least 89 countries: send students home with a crying, pooping, burping doll to teach them first-hand lessons of early parenthood.

But a surprising new Australian study has found that infant simulation dolls may actually increase the likelihood of pregnancy among teenage girls, not lessen it.

The research, published Wednesday in The Lancet, studied more than 1,000 teenage girls who went through Australia’s Virtual Infant Parenting (VIP) program, which is adapted from the U.S. program Reality Works – sometimes called the “baby think it over” program.

The VIP program is an opt-in curriculum that touches on issues like smoking, drinking and drugs and includes an infant simulator that students take home for a weekend. The doll cries when it becomes “hungry” and requires students to rock it to sleep, change it and burp it. A small computer inside the doll also measures any mishandling, crying time and overall care.

To gauge the program’s effectiveness, researchers compared 1,267 girls who went through the six-day VIP program to 1,567 girls who received a standardized health education curriculum. All participants were between 13 to 15 years old and lived in Western Australia at the start of the study.

Researchers then tracked each participant until the age of 20 using data from hospital records and abortion clinics.

The results showed that girls who cared for a baby simulator had higher rates of pregnancy and abortion than those who did not. About 8 per cent of girls in the VIP program (97 girls in total) had at least one birth by 20, versus 4 per cent (67 girls) in the control group.

A similar discrepancy was found in abortions. About 9 per cent (113 girls) in the VIP program had an abortion compared to 6 per cent (101 girls) in the control group.

The study’s lead author said these results are a warning sign for educators using the dolls.

“Similar programmes are increasingly being offered in schools around the world, and evidence now suggests they do not have the desired long-term effect of reducing teenage pregnancy,” wrote Dr Sally Brinkman of the University of Western Australia in a statement. “These interventions are likely to be an ineffective use of public resources for pregnancy prevention."

The authors pointed out that, while the study considered a large sample of teenage girls, the rate of participation in the study hovered around 50 per cent.

Researchers also highlighted that the VIP program is voluntary in Australia, and that girls in the control group had a higher average socio-economic status than those in the VIP program. However, researchers say these factors did not change the results.

The study has been called the first randomized controlled trial to test the effectiveness of infant simulation dolls.