Slightly different visualisations of Covid data
Typically, a country’s daily confirmed coronavirus cases and deaths are visualised in a manner similar to the graph below. Time is shown to progress along the x-axis, while the daily increases or decreases in cases or deaths would jump up or down along the y-axis. This makes for a compelling visual, and can be easily paired with a similar chart showing ‘cumulative cases’. These charts also help reinforce the message of ‘flattening the curve’, allowing people to easily understand the progression of the disease and how well health services and public safety guidelines may be helping in containing the spread.
I have used data from Public Health England to recreate such a daily cases curve for the UK. These data help show the number of people with at least one positive COVID-19 test result (either lab-reported or lateral flow device), by specimen date.
It should be noted that for tests conducted until at least July, there is a risk of cases being underreported. The initial lack of scaled testing means that many Covid cases would not be identified in these data.
I have combined the UK government data on case numbers (proxied here by the number of confirmed positive tests) with information from the Health Foundation, who tracked the timeline of national policy in response to Covid-19.
Using these 2 sources of information, I have arbitrarily assigned to the UK 3 waves of the Covid-19 pandemic. This is based on my own judgement and the strength of containment strategies. As shown in the figure below, I judge the first wave to have lasted between March and August; the second between August and late November; and the third starting in late November. These correspond to the rising and falling case numbers, but may not strictly fit into official classifications of waves.
Uniformity can be very good to establish a firm message, especially in trying times when the public needs to get a good grasp of the most recent covid numbers. If we all see the same curve, spreading the ‘flatten the curve’ message becomes easier.
I should say, at this point, that what I present below is not an attempt at serious epidemiology or public health work. It’s simply an exercise in data visualisation and an attempt to find creative ways to present information about Covid-19.
The first thing I’ve experimented with is a ridge plot (sometimes called a Joyplot because it appeared on an album cover by the band Joy Division). In this chart, time progresses top to bottom (oldest data at the top, latest at the bottom), and daily case numbers increase left to right. Each dot in the chart below is one day in the corresponding month, placed horizontally based on the number of confirmed cases on that day (as on the x-axis). So in the first wave (in green), many days cluster closer to zero. This is partly a result of insufficient testing, but assuming some consistency and improvements in testing, we see the first lockdown being somewhat beneficial. In the second and third waves (orange and purple) things get much worse with the number of cases per day spreading out significantly. While this chart doesn't help with the ‘flatten the curve’ narrative (*insert meme about the UK flattening the curve along the y-axis*), it can help with another potential public health slogan: “Stop the spread”. ‘Good’ months have more days clustered together near the left, and ‘bad’ months see a more spread out distribution.
I have repeated the above curve with the daily death statistics, the data for which can be found here. While there were early problems with testing in the UK, reducing the reliability of those data, the same is less true with death statistics. The ridgeplot below probably represents the UK Covid-19 situation better than the the plot for cases did.
Moving away from a within-country analysis to a cross-county comparison, a heatmap might help compare the status of various countries struggles against Covid-19. Line charts work because they make it easy to relate a specific number with a specific point in time. If data are to be aggregated and presented at a higher level and in a more general fashion, there needs to be some consistency in the metrics used and in data availability.
I have used data available from the organisation Our World in Data who have done excellent work related to Covid and beyond. Please follow their work and donate if you can.
Below is a heatmap of the total monthly cases (and deaths) in a country per million people (another calculation can be cases adjusted for population). This makes it easy to compare different countries in one glance. Data are available consistency across countries only for the months between March and December.
I am far from an expert on public health and do not want to assume my armchair analysis to be the best way of thinking about or presenting these data. This is, in the end, an exercise in passing along the message within the Covid-19 statistics in a slightly different way. I hope it was at least partially successful!
