Social network analysis: quirks, pitfalls and biases

This is the latest in a series of blogs exploring the use of social media in public health and healthcare. The blogs have used social network analysis to study awareness raising campaigns (#VaccinesWork, National Clean Air Day, Antibiotic  Awareness Week 2016), conferences (European Public Health conference 2016, Quality2017), and key influencers (exploring whether the 85:3% rule applies to tweeting about health and healthcare).

You can also read and download the blog in a PDF.

This blog attempts to share some key methodological pitfalls in planning, conducting and sharing the results of a social network analysis. It is structured around 5 main ideas:

  1. Finding a needle in a haystack
  2. Filtering out the minnows and sticklebacks
  3. Working out the size of pond for the big fish
  4. Slicing out the spam
  5. The ones that got away – and how to include them in the final analysis

The work has reminded me of work on cell culture when I was a medical student: I worked in a lab in Dallas, Texas, for 4 months November 1994-February 1995, studying adrenal tumour cells. This was my first time working abroad, and I was a little star struck, working in Parkland Hospital, famous as the hospital that treated JFK on 22 November 1963, and home to Nobel Prize winners including Alfred Gilman. My research did not reach such heights. Results were disappointing and unpredictable, week after week, and I was running out of time. Research is marked as much by its “failures” as its “successes”; both are an essential part of learning, though the stumbles are shared less than the leaps forward.

Continue reading “Social network analysis: quirks, pitfalls and biases”

Some thoughts on digital innovations in Public Health

Between 25 and 28 January 2018 thirty teams of healthcare workers, tech specialists and entrepreneurs will work together to develop new digital innovations in Public Health in a Product Forge hackathon. I have been asked to provide some thoughts in introduction. Preparation will be important for all participants, so here is my initial advice under 4 headings.

Continue reading “Some thoughts on digital innovations in Public Health”

#EuroTestWeek and #HIVTestWeek

During November 2017 there was a focus on HIV testing across Europe:

Both campaigns asked me to map the tweets over the course of the week. I am not associated with or funded by either campaign. I am doing this out of interest, as a Public Health doctor, in personal time. As for previous analyses written up on the ScotPublicHealth blog I am interested to see how different campaigns unfold. Of particular interest here is how tweeters observe the geographical “boundaries” of the campaigns – for UK, Europe, worldwide. I am also interested to see whether the tweeters and tweet content differ between the different campaigns and across the week. As ever, the ideal tweet grabs attention (image), shares information (in image and text, but also linking to further information using  a URL), and links to others (by mentioning other users/ tagging others in an image, and a hashtag).

I looked at the campaigns in a number of ways:

  1. Simple counts of Twitter activity (number of tweeters, tweets, estimated views of these tweets) – the hashtags are under review by the Symplur healthcare hashtags website but if accepted information will be available for the full period of the campaign. Followthehashtag also provides estimates, breaking “tweets” down into tweets, replies and retweeter, plus an estimate of “audience”.
  2. Geographical maps of tweeters/ retweeters using the Followthehashtag website: to look at spread of the message across the world (limit typically 1,500 tweets so will only be able to capture snapshots of larger campaigns).
  3. Social network analysis using NodeXL, looking at connections (people tweeting using the hashtags, retweeting or replying to these tweets, or mentioned in the tweets). This tool extracts up to 18,000 tweets and it is possible to combine extracts to look at longer period.
  4. Summarising top content using my “top tweets” analysis of NodeXL extracts – see examples of this approach on my Storify page.

Continue reading “#EuroTestWeek and #HIVTestWeek”

Hairmyres Hospital Allergy Awareness Week, 13th -17th Nov 2017: #AllergyHM17

This is the first “guest blog” on the ScotPublicHealth site. Fran Kerr, Antimicrobial Pharmacist, provides the background to Hairmyres Hospital Allergy Awareness Week (#AllergyHM17), both in the hospital and on social media. Social media metrics (NodeXL map, and other tools that Fran has access to) will follow during the week.

If you’d like to contribute a blog to the ScotPublicHealth site then please get in touch with Dr. Graham Mackenzie, Consultant in Public Health Medicine.

The aim of Hairmyres Hospital Allergy Awareness Week (#AllergyHM17) is to improve assessment and documentation of patient medication allergies to reduce harm associated with the administration of medication to which the patient is allergic.

figure 1
Daily ‘gold star ‘medals

Continue reading “Hairmyres Hospital Allergy Awareness Week, 13th -17th Nov 2017: #AllergyHM17”

World Antibiotic Awareness Week 2017

World Antibiotic Awareness Week 2017 runs 13-19 November 2017. European Antibiotic Awareness Day in on 18 November every year. Antibiotic resistance and antibiotic awareness are important public health topics. The World Health Organisation lists prevention and control measures for different sectors including individuals, policy makers, health professionals, healthcare industry and agriculture sector.

As previously discussed on this page antibiotic awareness campaigns attract a lot of tweets – professional and public facing. However it’s not clear which hashtag to use to tweet, or to search for during the week. This blog will be updated throughout the week to help you decide which hashtags to use and search for. I will summarise the top content using “big data” methods, listing the most retweeted tweets across the different hashtags.

The weekend before the campaign I ran a series of “follow the hashtag” searches (to map tweets geographically, and estimate audience) for different hashtags, and three large NodeXL social network maps. Links to these analyses, and top tweets from the NodeXL analyses are provided below. I also produced a graphic of my own to provide a personal touch that I could also track during the course of the week.

Final graphic for waaw 2017 (text tidied up_upload)

Continue reading “World Antibiotic Awareness Week 2017”

Who are the top 3% of influencers on health/ healthcare on Twitter? (work in progress)

Introduction: ScotPublicHealth work has always had a double purpose – exploring both public health and social media techniques. This post uses social media – and social network analysis (using NodeXL) in particular – to document tweeting about health/ healthcare. It looks at top influencers at UK and global level.

I have developed an interest in social network analysis over the past year. The method allows us to look at interactions between Twitter users. It also provides a complete extract of tweets between stated time limits that can be used for further analysis in Excel.

Back in July 2017 Helen Bevan‘s team at NHS Horizons contacted me to see whether these methods could be used to identify the top 3% influencers around health/ healthcare tweets. I set out to refine the social network analysis methods that I had been applying to simple Twitter searches (eg around a public health awareness campaign or conference, with tweeting around a single hashtag) to answer this rather more complex question.

3 85 question

There were 3 main purposes to this work:

Purpose Uses
To attempt to answer the 85%:3% question and identify the key influencers Help these tweeters understand their reach and influence

Identify other influential health/ healthcare community who may be creating useful content

To refine and describe the methodology Allow others to repeat and refine the approach and apply to other questions

To allow peer review

To summarise the type of content identified To understand the type of content that is achieving success, and the tweeters generating that content

Continue reading “Who are the top 3% of influencers on health/ healthcare on Twitter? (work in progress)”

What healthcare workers can learn from Twitter via green spaghetti junctions

The BMJ cover story on 9 September 2017 described the response to a Twitter ban at a large conference in the US. Was this a lot of witter, a social media storm in a teacup; or did it represent censorship and a missed opportunity to learn at a conference?

(Prefer this blog as a PDF? Download it here)

Source: BMJ cover story, 9 September 2017

I started using Twitter in 2013 when “early adopters” were evangelising about its potential. I learnt by observing others and started tweeting. Following people with shared interests led to connections and further learning. However, the idea of scrolling through lots of tweets on a small device has never appealed very much.

Twitter provides a glimpse of clinical and scientific developments, often well ahead of print. Colleagues share ideas in accessible ways (eg infographics, short films, or blogs). However, the volume of information quickly becomes overwhelming. Big names across clinical specialties are on Twitter. Click from person to person, through hashtags, follow Twitter recommendations and you’ve lost an hour without much to show for it. You need a strategy to use Twitter effectively.

Conferences are a great way to start using Twitter in a more structured way. Delegates and conference organisers often start tweeting using the conference hashtag a few days in advance. You will arrive at the conference prepared, with ideas of work to search out and people to meet. However, after the conference has finished you need to have a plan about how to stay in touch and informed.

As a Public Health doctor I am interested in the big picture. The first time that I saw a social network analysis (SNA) map (of a conference via Mark Outhwaite, July 2015), I was hooked. These maps show interactions around a Twitter search term. It took me over a year to try my own maps, and another few months to understand the potential of these mystifying and beguiling green spaghetti junctions.

figure 2

Source: NodeXL social network analysis of the 5 top global medical journals, 19-26 September 2017

In October 2016 I started to use SNA tool NodeXL to work through a series of topics, including conferences and health campaigns. Automated reports provide summaries including top influencers, hashtags and weblinks from tweets. However, richer information is available from the individual tweets captured by NodeXL. Sifting through tens of thousands of tweets takes a few minutes using approaches I have refined over recent months. I identify the most shared posts as a rapid means of “digital curation”: if something has been shared a lot it is either popular or controversial, so there will be something to learn. This approach gives a contemporary glimpse into the response of professionals, patients and/or public. You can access the resulting “big data” summaries even if you don’t tweet: e.g. see my summary of BMJ/ IHI International Forum on Quality and Safety in Healthcare #Quality2017.

The approaches have wider application. Studying Immunization Week 24-31 April 2017 (hashtag: #VaccinesWork which is used year-round) I was interested – and concerned – that one of the tweeters with the biggest potential audience (number of followers multiplied by number of tweets(*)) was an anti-vaccination campaigner. Some weeks his tweets had a larger potential audience than large global campaigning organisations such as the World Health Organisation and UNICEF. I used SNA to study his connections. The anti-vaccination campaigner might have had a lot of followers, and a lot of potential views, but his tweets were hardly connecting. He was howling into the void. In comparison, the global organisations, and many individual vaccine advocates, achieved a much larger number of connections with accessible, clear and evidence-based materials explaining the benefits of vaccination for a general audience. The full analysis is available on my blog.

(*) Go to “get analytics” at bottom of Symplur page and set dates you want to study.

Source: #VaccinesWork tweets using NodeXL.

In addition to searching for hashtags it is also possible to use NodeXL to study people, organisations, weblinks and words and phrases on Twitter. This approach can be deployed to look at specific questions – for example to study the considerable social media impact of the BMJ cover story on the conference Twitter ban. The method is also useful in keeping up to date with top medical journals more generally. A recent search (19-26 September 2017) identifies top content from almost 23,000 tweets by over 13,000 tweeters. The resulting top 50 summary includes videos of clinical examination (including an ocular trematode featured in NEJM), clinicians’ responses to journal articles and tightly worded summaries, with weblinks to clinical and academic content from across these journals.

Julio Mayol, an early adopter of Twitter in his work as Professor of Surgery, writes about Twitter and surgical research in the December 2017 issue of BJS, but his points refer equally to surgery and clinical work more generally (see list below). The issue’s front cover features one of my NodeXL maps, after being approached by the journal’s social media champion, surgeon Richard Brady on 15 October.  As captured in a subsequent NodeXL map, BJS and Julio’s tweets about the article and journal issue feature prominently in tweets by and about surgical journals in November. Indeed 11 tweeters in that more recent NodeXL map achieve 85% of the retweets on that map (out of 146 tweeters). The speed of publication, and the journal’s interest in social media, social networks, and infographics, is impressive, and is amplified by the work of individual surgeons like Julio and Richard.


Twitter therefore has huge potential for healthcare workers for CPD and networking, and to study their impact if active on Twitter. However, you don’t even need to be on Twitter to access this content. The approach to digital curation described here can provide an up-to-date view of the medical and scientific world, driven by and feeding an endless curiosity and desire to learn. To see this work moving from niche interest/ hobby to the front cover of two global publications (BMJ and BJS) over a period of 3 months is exciting, and the interest looks set to continue with a paper on my work on conference tweeting in press for January 2018. Meanwhile, I have stopped actively tweeting, concentrating instead on finishing my studies of health and healthcare tweeting without influencing the topics that I am studying.

Dr Graham Mackenzie, Consultant in Public Health, NHS Lothian, 6 November 2017; updated with paragraph on Julio Mayol and BJS, and the final two sentences of article, on 19 November


Links: My further outputs from NodeXL analyses are available at my ScotPublicHealth blog and Storify page, and these are updated regularly.

I have no conflicts of interest to declare. This work has not been funded externally. I would like to thank Marc Smith and his colleagues at SMRF for their help in getting NodeXL set up on my computer and their ongoing input in refining and developing their software and helping me improve my analyses.

Thanks to Julio Mayol for his kind words about this blog:

Julio Mayol tweet

What’s in a hashtag? An analysis of tweets posted during World Antibiotic Awareness Week 14-20 November 2016

Antimicrobial resistance presents a major challenge, but simple steps can to minimise the impact. Social media has a role in disseminating information during antibiotic awareness campaigns: information can be shared rapidly, widely, and cheaply.

Hashtags are used to identify social media posts on a specific topic. The official hashtag for social media activity during the 2016 world antibiotic awareness week (14 to 20 November 2016) was #AntibioticResistance, with a separate hashtag for the official European campaign on 18 November 2016 (#EAAD or #EAAD2016).

However, numerous other hashtags were used quite widely during the week, including hashtags for established professional-focused national campaigns such as #AntibioticGuardian, which originated in the United Kingdom (UK), and an unofficial hashtag that reflected the purpose of the week – #AntibioticAwarenessWeek. There was also some tweeting without hashtags. There were many tens of thousands of tweets posted during the campaign week by tens of thousands of individuals and organisations.

It would not be possible to analyse this volume of tweeting using conventional methods. Accordingly, we extracted data using social network analysis (NodeXL) and Twitter searches to study the characteristics of tweeters and tweets using the most frequently used hashtags (official and unofficial) and without hashtags. An example NodeXL map is shown in Figure 1.

The quality of the most popular tweets was assessed by looking at use of hashtag, image or mention of another tweeter. These are components that boost the impact of a tweet as described in the previous ScotPublicHealth blog post.

The aim of the analysis was to identify whether there were differences in tweet content between the hashtags, to see if there is value in looking beyond the official hashtag, both for learning, networking and broadcasting.

Continue reading “What’s in a hashtag? An analysis of tweets posted during World Antibiotic Awareness Week 14-20 November 2016”

Increasing the quality and impact of Public Health tweeting – looking beyond #ScotPublicHealth

I’m writing this blog in preparation for the European Public Health conference in Stockholm, 1-4 November 2017 (#EPHStockholm). At the conference I will be presenting the #ScotPublicHealth work on quality and impact of tweeting during 2016 (read JPH paper here; please contact me if you do not have access to the article and would like a copy). I have recorded a dry run as a “video abstract” (far from perfect, but useful preparation for the presentation itself). Full slides are available too.

The #ScotPublicHealth work emerged from the Scottish Public Health conference in Peebles, November 2015. It was clear that we should be attempting to network and share better across the Public Health community in Scotland – it was not enough to meet at an annual conference. Colleagues spread across large distances in Scotland did not have an opportunity to meet and discuss topics of mutual interest. I set out to improve connections using Twitter, but quickly learnt that colleagues wanted more advice about tweeting (see write up of #ScotPublicHealth tweet chat January 2016).


The series of blogs and webinars that we ran around #ScotPublicHealth in 2016 has been described in a series of blogs on this page; these activities were an attempt to learn more about public health topics and social media and internet technology. We looked at physical activity with Prof Chris Oliver, assets based approaches to improving health with Glasgow Centre for Population Health, health inequalities with NHS Health Scotland and Institute of Health Equity, and realistic medicine with Scotland’s Chief Medical Officer and an expert panel.

I compared tweets posted about “public health” in Scotland between September and December 2014 and compared these with tweets using the #ScotPublicHealth hashtag Autumn 2016. There was an increase in the quality of tweets as assessed by the % tweets with an image, URL and/or mention.


There was also an increase in the impact of these tweets as assessed by the % retweeted (and the number of retweets achieved). The greatest impact was seen in tweets that used more than one key component (mention of another Twitter user, image and/or URL).


It is clear from the findings that while general “public health” tweeting has improved, conference tweeting remains an area for improvement. Some people will only tweet on professional matters when prompted, for example at conferences. In my study this group of “conference tweeters” was less experienced than colleagues who had engaged in the #ScotPublicHealth work throughout 2016. Conference tweets in the #ScotPublicHealth study were less likely to include an image (which draws the eye in), a URL (which provides further information), or mention another Twitter user (which helps sharing). There was only a modest increase in quality of tweets between the 2014, 2015 and 2016 conference (⬆pictures, very small increase in use of URL though remained a small minority at 12% in 2016; no increase in mentions of other Twitter users).

The low use of URLs in conference tweets is a feature of other Public Health conferences – eg FPH conference in Telford July 2017EPHVienna conference November 2016 and  APHA conference November 2016. The under-utilisation of URLs at Public Health conferences is a missed opportunity. Microbiologist Jon Otter and colleagues have demonstrated that the URL is important in professional sharing, and this has resulted in very effective and informative tweeting at microbiology conferences (paper in press). If there isn’t a link to a paper, blog or video that explains the findings presented at a conference then people will be less likely to share it on social media. As scientists and clinicians we need to assess the credibility of information before we can assess whether we want to broadcast it.

Continue reading “Increasing the quality and impact of Public Health tweeting – looking beyond #ScotPublicHealth”