From the Eurostat website you can download a lot of different datasets to play with. I got my hands in a dataset about health care expenditure by financing agent to see how health care is financed in some countries, actually all the countries from the dataset, which are all EU 27 (excluding EL, IE, IT, MT, and UK), Norway, Iceland, Switzerland, Japan, USA, Australia and Korea. The data shows how is health care financing by different sources (with subdivisions) and by different parameters. More information about the dataset can be found here.

I will see first how is the mix from two different main sources, general government and private sector. I will see what percentage of each makes the funding of health care in the different countries and how it changed over the last years.



Only the US and Cyprus fund their health care system with more money from the private sector than from the general government. There are some interesting changes over the last years and you can see the effect of some law changes concerning health care funding in several countries. 

I will plot now the total health care funding, the sum of all financing agents, as a percentage of the GDP over time.



We can see that most countries, but not all, are slightly increasing funding for their health care systems.

Finally I will rank the countries by total funding as a percentage of the GDP. Since there is no data for 2012 for all the countries I will take the data from the last year that has data available.



You can find the code to get the data and reproduce the graphics here.



For today's graphic I am going to use the data from table 3 of the article United Kingdom National Ophthalmology Database Study of Vitreoretinal Surgery: Report 1; Case mix, complications, and cataract. In that table are given the reported intraoperative complications and I am interested in the complications happened with a pars plana vitrectomy (PPV).

I will plot the complications rate but will add a 95 % confidence interval for the true rate.



The number in black is the percentage rate, the green one is the lower confidence interval and the red one is the upper confidence interval. As there are 25 complications, there will be at least one that is out of that interval. Any guesses?

You can find the R code to reproduce the graphic here.


After the last post, some people asked me about the trends of more refractive surgery terms. Since there is some interest I will take a different approach and I will put the terms from last post and some more in a word cloud. The bigger the word, the more articles in PubMed that have that term in their title or abstract for 2014, up to 29 November.



The refractieve error terms are the most frequent, as expected, and then the more general terms like Lasik, PRK or Refractive surgery. Now let´s see which one of those terms had the largest rate of articles in the last four years.



The size of the term is now related to the maximum deviation of the rate of that term from the mean of the four years and the color represents the year where that maximum happened. So although a term might not have the largest amount of articles in total with that term in it, it can have the highest rate since that year there were less articles for all the terms. That happens for example with the PRK term, the maximum amount was actually in 2012 with 92 and for 2014 was 82, you can check it in the graph from the previous post, but the amount of articles with all the terms in 2012 was 2000 and in 2014 was 1733 so the rate for 2012 is 92/2000 = 0.046 and for 2014 the rate was 82/1733 = 0.047 so in the comparison cloud PRK goes to 2014.

This explanation is based in the documentation of the worldcloud package for R that I used to make it.

You can find the code to make the graphs here but although the size of the words will be the same, the place of them is different every time you run it.