A new data analysis, published in Vaccine has found that women and menstruating people who had one or more mRNA-based COVID-19 vaccinations were more likely to experience changes to their menstrual cycle in the six weeks following their vaccine.

The changes, detected using machine learning, were transient and lasted up to three months. 

The study from Murdoch Children’s Research Institute (MCRI) interrogated a large Australian primary care dataset of de-identified information from people in Victoria and New South Wales who presented to local GPs with concerns about menstrual cycle changes.

MCRI's Dr Aishwarya Shetty said the study used machine learning to first identify symptoms raised on social media by women and people who menstruate, who felt that their concerns were not listened to.

“We used machine learning to detect an increase of social media posts about changes to periods following COVID-19 vaccination – this was then validated through our analysis of general practice data,” Dr Shetty said.

Image: Dr Aishwarya Shetty

“This research underlines the importance of listening to community-level concerns about public health interventions, including vaccines, and the need to validate and address them to help curb vaccine hesitancy.”

Using data from social platforms like Reddit and primary health information from Australian GP clinics made the analysis more robust than similar international studies. These studies used self-reported data on menstrual tracking apps and post-vaccination complaints forms, both of which carry inherent bias.

“Our data was taken from people who went to their doctor with a menstrual cycle concern they may or may not have associated with a COVID-19 vaccination - and the analysis we performed was able to control factors that other studies couldn’t,” Dr Shetty said.

Dr Shetty said that machine learning technology could also be used to identify community concerns much faster than traditional methods and could routinely monitor concerns within, or beyond, women’s health.

“Our established system VaxPulse proved effective in this study as it could in future monitor social media to alert us to new outbreaks, medication side-effects or public health concerns and we could investigate it using large datasets within days or even hours” she said.

Dr Shetty said her team would aim to develop new strategies to respond to community concerns in an effective way.

“We listened, and we will continue to listen to all your vaccine-related concerns,” she said.

“This has been a valuable body of work and shows MCRI could play a key role in tackling public health concerns faster and more effectively while enabling researchers to listen, investigate and inform across consumer health and health informatics.”

Not-for-profit organisation POLAR by Outcome Health also contributed to the findings.

Publication: Aishwarya Shetty, Gonzalo Sepulveda Kattan, Muhammad Javed, Christopher Pearce, Jim Buttery, Hazel J Clothier. ‘Validating community concerns of menstrual changes associated with COVID-19 vaccination using a self-controlled case series analysis of real-world data,’ Vaccine. DOI: 10.1016/j.vaccine.2025.127511

Available for interview

Dr Aishwarya Shetty, MCRI Epidemiologist, Epidemiology Informatics
Professor Jim Buttery MCRI Group Leader, Epidemiology Informatics
Associate Professor Hazel Clothier, MCRI Lead Epidemiologist, Child Health Informatics and Surveillance

Media Contact

Murdoch Children's Research Institute
Phone: +61 457 365 848
Email: show email address

About Murdoch Children’s Research Institute

Murdoch Children's Research Institute (MCRI) is the largest child health research institute in Australia committed to making discoveries and developing treatments to improve child and adolescent health in Australia and around the world. They are pioneering new treatments, trialling better vaccines and improving ways of diagnosing and helping sick babies, children and adolescents. It is one of the only research institutes in Australia to offer genetic testing to find answers for families of children with previously undiagnosed conditions.

Funding

The work was funded by the Department of Health, Victoria. They were not involved in collection, analysis, interpretation of data; writing of the report; and in the decision to submit the paper for publication.

We thank Mr. Adam McLeod and Karina Gardner of Outcome Health, Melbourne, Victoria, Australia for their support with this research and Dr Diana Vlasenko (Epidemiology Informatics, Centre for Health Analytics, Melbourne Children’s Campus, Victoria, Australia) for assistance with mapping of SNOMED CT codes. We also thank Dr. Gerardo Luis Dimaguila for conceiving the VaxPulse project and for his valuable contributions to this work. They did not receive any compensation for these roles.