How we built FluTracking

Flutracking logo

Craig Dalton and Sandra Carlson

FluTracking is an online weekly survey of respiratory symptoms that has tracked influenza from 2006 and then COVID-19 during the pandemic. It had a peak weekly participant count of over 150,000 people across Australia and New Zealand in 2021 and has expanded to Hong Kong and Argentina.  This post briefly summarises its genesis and development.

In 2005 we read an article “Did you have the flu last week”, a single telephone survey of influenza symptoms in Sweden. It seemed this point prevalence survey, while useful, was limited by being a single week and it was clearly something that could be repeated at minimal cost with a weekly email-based survey.  While there were off the shelf online survey packages available, none of them at that time were designed for managing an ongoing cohort with flexible question options based on prior or current week responses. We had to build FluTracking from the ground up.  As the developer finished his last line of code, after many frustrating testing iterations, he pushed back from the desk and declared, “I have just wasted three months of my life, no one will answer this stupid bloody survey.”

We were not sure he was wrong; why would people answer a survey about flu symptoms every week for 24 weeks? After emailing approximately 7,000 email addresses in 2006 in the Hunter New England Health District, the very first FluTrackers joined – some of whom are still participating in the surveys in 2022!

The survey responses began to slowly come in, and over the winter of 2006 we had a total of 400 participants by the end of the survey season in October. Four hundred was great as a proof of principle but we knew we needed many more to begin understanding respiratory symptom rates in rural and metropolitan areas, across a wide range of ages, and among Aboriginal and Torres Strait Islander people. We dreamed of having 5,000 at some point in the future and we thought the maximum we could ever recruit would be 25,000. In 2008, we tweaked the survey design to allow participants to respond to the survey on behalf of other household members; this boosted participation by around 60%, and allowed us to capture data on children.

The largest increases in participation have come with events such as the 2009 H1N1 influenza pandemic and the 2020 COVID-19 pandemic. Our most successful recruitment campaign has always been an email to current participants asking them to invite their friends. We split test our email invitations as mini-experiments to understand which ones produce the greatest number of new recruits. Our experiments showed that it was better to ask people to “invite five friends” rather than just “invite your friends”.  We then split test “five friends” versus “three friends”, which won, and then tested “two friends” which had even higher recruitment success. Informal discussion with participants revealed that insights such as “when you asked me to invite five friends, it was kind of a big task and I was paralysed, [but] when you asked me to invite just two friends I was able to think about who would be the best two people to do the survey.”  In 2021, the “ask two friends” email recruited over 6,000 new participants.  Another major recruitment came when the then Deputy Chief Medical Officer Paul Kelly asked Australians to join FluTracking during a live media conference in 2020, resulting in over 20,000 new recruits the same week.

As FluTracking has always been built for flexibility, we were able to track smoke-related symptoms during the devastating summer bushfires of 2019/20 and mount a survey of thunderstorm-related asthma in Victoria in 2020 in a matter of days.

FluTracking provides another perspective to that of traditional systems; however, it is not perfect and we need to constantly triangulate data between the different surveillance systems. One of its strengths is that it doesn’t send an exaggerated signal in the way that emergency department, general practice, or laboratory systems sometimes do when there is community panic or increased testing practices. But at the same time, because only a small percentage of FluTracking participants have self-reported laboratory confirmation, we need to use the other systems to provide a coherent picture of virological incidence of disease.

In the early phase of the current pandemic, FluTracking provided evidence of a significant decline in community level respiratory illness rates in Australia coinciding with initiation of public health measures against COVID-19 and preceding the rapid decline in COVID-19 cases in Australia.

In late 2021/early 2022 FluTracking picked up the first and second wave of Omicron attributed to the BA.1 and BA.2 variants respectively, and identified the peak of each wave in real time. FluTracking also showed that prior to the introduction of rapid antigen tests (RATs), only 50% of people with new fever and cough symptoms, and 30% of people with sore throat and runny nose symptoms were reporting being tested for COVID-19. Once RATs were introduced, and the Omicron variant took hold in Australia, the rate of testing via both PCR and RATs increased to over 90% for participants with fever and cough.

The FluTracking team will be monitoring closely to see how influenza plays out over the 2022 winter, after two years of very little transmission.

When we ask FluTrackers why they do the survey, they often find it hard to explain. Typically they say something like, “It only takes 20 seconds on a Monday morning and I feel like I have done something useful for medical research.”

If you have creative ideas on how to promote FluTracking to friends, colleagues or organisations, we are open to trying new methods.

Anyone in the community can join FluTracking (sick or healthy/vaccinated or unvaccinated) and answer a simple 20 second survey each week about their respiratory symptoms – please send your friends and colleagues to to join, or retweet @Flutrack.

Craig Dalton is a Conjoint Associate Professor at the School of Medicine and Public Health, University of Newcastle, and a Public Health Physician at Hunter New England Population Health. Sandra Carlson is the Flutracking Manager, Health Protection at Hunter New England Population Health.


Leave a Reply