Mozzie bites and tweet tracking

One of my favourite artists, Nat Russell, painted a wonderful portrait of me a couple of years ago

One of my favourite artists, Nat Russell, painted a wonderful portrait of me a couple of years ago. Perhaps think of this as me wading out into the sea of social media?

Could social media help beat the bite of mosquito-borne disease?

Social media won’t do it alone but I think it is definitely something Australian authorities should embrace. The only problem is, how do you measure the success of social media activity? Taking my activity on Twitter as a case study, I monitored the changes in follower number, “tweet type” and estimated reach and exposure of tweets over a six month period. This was during a time when I would normally be active in the media responding to  mosquito-borne disease outbreaks or general interest questions about mosquito biology.

As broad scale mosquito control programs are generally limited, Australian health authorities typically rely on the communication of personal protection strategies to reduce mosquito-borne disease risks. These personal protection strategies may include avoiding known mosquito habitats, wearing long sleeved shirts and long pants to create physical barriers to biting mosquitoes and the use of insect repellents. Messages are usually relayed to the public via media releases or online fact sheets.

I started using Twitter in September 2010 with the expectation that I could use the service to distribute those public health messages as well as news on mosquito and mosquito-borne disease research. I generally tweet material that is related to my position with NSW Health/Westmead Hospital/University of Sydney but my account is not an official source of information from those organisations. I generally keep “personal” tweets to a minimum.

I’d already had some experience with public health communications working groups. I consider my activity on Twitter to be an extension of that work. In particular, my work with the “Living with Mosquitoes” group in the Hunter region investigated new ways to raise awareness of mosquito-borne disease risk and communicate more effectively the benefits of personal protection strategies. A couple of the options we tried were the incorporation of “mosquito risk periods” into free tide charts and stickers designed for primary school students. We even briefly (unsuccessfully) experimented with using myspace to host some information.

Using Twitter to spread the message

How did I go about using Twitter to help spread the word on mosquito-borne disease? At first I was expecting to build a following directly with the public by growing the number of followers. What I’ve found, however, is that the greatest benefit of Twitter has been when it is used in association with traditional media activities. Tweets can be exchanged between myself and the presenter/broadcaster/publisher, particularly links to online resources/fact sheets, and then subsequently retweeted to their followers.

You can read more background about my use of Twitter for spreading mosquito-borne disease awareness in this article, “Can the buzz of mosquitoes be replaced with a tweet?”, recently published in “Mosquito Bites” – the newsletter of the Mosquito Control Association of Australia.

While I initially thought a large number of followers was important, I now realise that engagement with the media (as well as other active users of Twitter) may be the best way to enhance the way health messages can be promoted. It is our local media that play the primary role in disseminating public health information to the local community, perhaps Twitter is best used to build communication lines between journalists, scientists and local authorities?

Assessing activity on Twitter

I started to think of ways I could better assess my use of Twitter to help answer some of these questions. I first starting thinking about this after reading a great paper by Thackery et al. in 2012 titled “”Adoption and use of social media among public health departments”. The paper describes the social media activity of health departments and highlights that very few use social media to engage the community. Their use of social media is, as is the case for traditional media, a one-way direction of information. There is very little active engagement. The authors argue that the departments need to develop a strategic communication plan to expand their reach while fostering interactivity and engagement.

This is very much the case in Australia too. If you have a quick look at state health department Twitter accounts (e.g. NSW Health, QLD Health), there is very little (if any) engagement with other Twitter accounts (i.e. very few RTs or Replies).

Before the start of the 2012-2013 “mosquito season”, I decided to try and document some of my activity on Twitter using some free online analytics services. I am the first to admit, this was a pretty rough and ready way to collect data. It was really just an experiment to see what kind of data could be collected to document how my activity on Twitter changed over the course of the season.

Most of the information was collected weekly from TweetReach. This website collects data on your account including estimated reach (total number of unique accounts that receive tweets) and exposure (total number of times tweets are received by any account) as well as a breakdown of “tweet type” (e.g. tweets, retweets and replies). It samples the last 50 of your tweets to collect this data. I logged in every Saturday morning and downloaded the data. I tweet more than 50 times a week (on average over this period I tweeted about 70 times per week) so the data represented what was going on towards the end of each week. I started in early November 2012 and stopped at the end of April 2013.

So, what did the analysis of my Twitter activity reveal?

Firstly, did my followers change over this time? There was a steady increase in the number of my followers as shown in the chart below. Followers increased from 916 to 1406 over the six month period. I’m not exactly sure what this reveals but since there were no notable falls in the number of followers, perhaps it suggests that most followers find the tweets of interest (or at least not annoying enough to “unfollow”).

A chart showing the weekly growth in my Twitter followers from November 2012 through April 2013

A chart showing the weekly growth in the number of followers from November 2012 through April 2013

Secondly, what did analysis of my “tweet types” show? There is generally a three way split in my activity between tweets, RT and replies. The trend remained fairly consistent over the six month period as shown in the chart below. Many of the RTs were tweets from various health authorities providing information on mosquito-borne disease outbreaks or other health related matter (e.g. infectious disease outbreaks, vaccination information, general health advice). It was generally a quiet season for mosquito-borne disease activity. The start of the season was marked by local activity of dengue in FNQ and the end of the season by activity of Ross River virus and Barmah Forest virus in SE QLD as well as SW WA. As a result, there was substantially less coverage of “mozzie stories” in the traditional media this season compared to previous years.

A chart showing the mix of "tweet types" in 50 of my weekly sampled tweets

A chart showing the mix of tweets, retweets and replies in 50 of my weekly sampled tweets

Many of the replies I tweeted were promoted by questions directed towards me following my tweets. Some were requests for more information or clarification on mosquito-borne disease activity or for my thoughts on recently published studies or news reports. I made an effort to respond to everyone who tweeted me. Interestingly, a recent paper by Neiger et al. (2013) titled “Evaluating social media’s capacity to develop engaged audiences in health promotion settings: Use of Twitter metrics as a case study” identified the number of questions and interaction between a user and their followers as a measure of “medium engagement”.

Finally, what was the reach and exposure of my tweets over this six month period? This was some of the most surprising information. Despite a relatively modest number of followers, my average weekly reach as approximately 19,000 and estimated exposure approximately 45,000. Much of the added reach and exposure came from multiple RT of my messages, not only accounts with large followers (e.g. media outlets) but from RTs by multiple users with similarly modest follower counts. I think this information shows the power of a small but engaged group of followers.

Chart showing the "reach" and "exposure" of the 50 tweets in my weekly sample

Chart showing the “reach” and “exposure” of the 50 tweets in my weekly sample

What influenced changes in reach and exposure?

Following the release of a health warning by NSW Health in mid-December 2012 (it is typical of health departments to release a warning about the start of the mosquito season every year) combined with a piece on mosquitoes on The Conversation, I was asked to do a series of radio interviews, mostly with stations in the ABC radio network. All had Twitter accounts that retweeted my link to repellent use guidelines following the interview. There would also often be some additional questions and comments tweeted about following the interview that I could respond to.

It is also easy to often forget who is following these accounts. In the case of 702 Sydney, whose account has over 20,000 followers, following an interview on why some people are bitten more by mosquitoes than others, I even received a tweet from the NSW Premier.

The peak in estimated exposure of my Twitter account came in early January. This was following an appearance on the Today show (a nationally broadcast tv breakfast show). Following an appearance to talk about mosquito repellents and their use, I had a tweet of mine retweeted by the producers and host of the program and this was subsequently retweeted by a number of their followers too. It provided exposure of a link to my guidelines for mosquito repellent use to almost 100,000 unique twitter accounts (with estimated exposure of approximately 188,000). As a result, I had over 200 visitors view the guidelines within a couple of days. That may be less than 1% of the people that saw the original link but still a substantial jump in the amount of people who would have otherwise visited the guidelines. I wonder how many people visit the “mosquito fact sheet” on the NSW Health website after a media release goes out?

So, what does all this mean for the potential benefits of Twitter?

In short, I think it Twitter provides a complementary route of community engagement to traditional methods. It certainly doesn’t replace any of the traditional methods of community or media engagement but I think it will become increasingly important in the future. From my experience, the ability to engage with local media outlets greatly increases the potential reach and exposure of information you can provide. This is particularly the case when links can be tweeted (and hopefully retweeted) that direct people to credible sources of public health information. The more people are aware of the risks associated with mosquito-borne disease, and the strategies available to reduce those risks, the better the public health outcomes.

The analysis of the reach and estimated exposure of my tweets demonstrates how, even from a Twitter account with a modest number of followers, messages can reach a much larger audience. That audience can be increased by being more engaged with followers. It isn’t just the Twitter accounts of media outlets and journalists. I’ve found that there are many active Twitter users who tweet and retweet material covering a wide range of topics. These users are actively engaged with a large cross section of other users and when they retweet material, messages are received by accounts that may not even think to seek out an account tweeting about mosquito-borne disease!

I am confident that the use of Twitter can assist in getting the community more engaged in public health issues, not only mosquito-borne diseases! Developing better strategies for the use of social media (by both health departments and individuals) as well as an assessment of whether those strategies are successful is required.