When: Wednesday 14th November, 2:40pm – 3:40pm
Where: L2, Level 2 to the left of the registration/foyer area, down the hallway and through the doors on the right
Hashtag: #W17

Background
Data visualisation is the broadly-applicable science of communicating information via graphical representations. The conventional technique for visualising drug dispensing over time is the run chart – a graph of data over time measured in a line. An alternative to the run chart is Nightingale’s rose diagram[1], which shows areas in a circle over time measured in a clockwise direction. No known studies have used Nightingale’s rose diagram to visualise drug dispensing temporally. Antibiotic use fluctuates with the seasons and, in modern society, is closely monitored over time because overuse enables infectious disease-causing bacteria to develop antibiotic resistance. In modern public health, it is important to visualise antibiotic use in a way that is understandable to health professionals, researchers, government officials, and the generally public. This study aimed to compare two techniques for visualising antibiotic dispensing over time: the run chart and Nightingale’s rose diagram.

Methods
Dispensing data were sourced from Pharmaceutical Benefits Schedule Item Reports[2]. For each month in 2017, data were extracted on number of systemic antibiotic (Anatomical Therapeutic Chemical 5 code J01[3]) prescriptions dispensed in Australia under the Pharmaceutical Benefits Scheme or the Repatriation Pharmaceutical Benefits Scheme. Antibiotic dispensing over time was visualised by producing run charts in Microsoft Excel 2013[4] and Nightingale’s rose diagrams in AnyChart[5], with stratification by drug class. These data visualisations were compared visually and with respect to elements ranked in order of decreasingly accurate perceptions of absolute quantity[6-8]: 1. position (common scale); 2. position (non-aligned scale); 3. length/direction/angle/slope; 4. area; 5. volume/density/curvature; 6. shading/colour saturation/colour hue.

Results
In the run charts and Nightingale rose diagrams, antibiotic dispensing increased from the lowest levels in January and February (the Australian summer) to peaks in August (the Australian winter). The investigator observed that Nightingale’s rose diagrams were eye catching and accentuated the seasonal component of the data. With regard to the accuracy of perceptions of absolute quantity, the run chart attained the highest score of 1 for position (common scale) whereas Nightingale’s rose diagrams scored lower: 3 for length and 4 for area.

Conclusions
When one is visualising antibiotic dispensing over time, choosing Nightingale’s rose diagram over the run chart gives accentuated seasonality but less accurate perception of absolute quantity. These two data visualisation techniques may complement one another. If one were to present these plots together or incorporate the run chart’s numeric scale into Nightingale’s rose diagram, then there could be improved communication of information on antibiotic use (or overuse) to health professionals, researchers, government officials, and the general public.

References
1. Magnello ME. Victorian statistical graphics and the iconography of Florence Nightingale’s polar area graph. British Society for the History of Mathematics Bulletin. 2012; 27: 13-37.

2. Australian Government Department of Human Services. Pharmaceutical Benefits Schedule Item Reports. Available at: http://medicarestatistics.humanservices.gov.au/statistics/pbs_item.jsp [Accessed 8 August 2018].

3. World Health Organization Collaborating Centre for Drug Statistics Methodology. ATC/DDD Index 2018. Available at: http://www.whocc.no/atc_ddd_index/ [Available at: 12 August 2018].

4. Microsoft Corp., Redmond, WA, USA.

5. AnyChart, St. Augustine, FL, USA.

6. Cleveland WS and McGill R. Graphical perception: Theory, experimentation, and application to the development of graphical methods. Journal of the American Statistical Association. 1984; 79(387): 531–554.

7. Cleveland WS and McGill R. Graphical perception and graphical methods for analyzing scientific data. Science. 1985; 229(4716): 828–833.

8. Shah P and Miyake A. The Cambridge handbook of visuospatial thinking. Cambridge University Press: Cambridge, 2005.

Session

Case studies: Visualising science

Presenter

Michael Leach, Adjunct Research Associate, School of Rural Health, Monash University