Nowadays quite a few health statistics are published online. Here I’ve got some graphs on some cancer stats published this year. The stats take a while to collect so we are only just onto the 2014 data (in the perfect world hospitals would submit a months stats the next month so the annual data could be published by the end of February. We are not living in that world yet. Also we wouldn’t get sick in the first place in that world, which is why it would be so easy to submit the stats).
This set of stats is pretty detailed, it lists the number of patients with each of the ‘big 13’ cancers, at stages 1, 2, 3, 4 or unknown for each CGG (clinical commissioning group) in England. Stage 1 is early and good, stage 4 is late and bad. The 1 year survival for the ‘big three’ (breast, bowel and lung) has also been published for each CCG (the % of patients diagnosed in 2013 who were still alive at the end of 2014). It’s interesting to compare these two dimensions by graphing early stage of diagnosis (by adding up all the patients diagnosed in 2013 and 2014, except three cancers are new with only one year of data) against 1 year survival for each cancer for each CCG.
I find it pretty interesting that three such distinct clusters just fall out of the data.
In the bottom left is lung cancer. This unfortunately has very bad early diagnosis with only 14-37% of patients diagnosed with stage 1 or 2 and equally bad 1 year survival. Of all the patients diagnosed in 2013 only 24-47% where alive at the end of 2014, depending on where in England they were.
Colorectal is intermediate, 30-57% of patients are diagnosed at stage 1 or 2 and 68-85% were still alive at the end of 2014.
Breast is best with 72-96% of patients are diagnosed at stage 1 or 2 and 93-97% were still alive at the end of 2014.
I think it’s interesting because the variation across the two dimensions for all the CCGs (each dot in each group) is smaller than the variation between the different cancers (the space between the groups). So which cancer you get will affect you more than any variation that exists around the country. Bear in mind that differences between CCGs, might be due to something about diagnosis/treatment OR they might be due something about the population. Deprivation definitely plays a role, for example deprived areas have a much lower screening, which will reduce early diagnosis. So do the graphs show variation in treatment or variation in populations? This isn’t exactly about lifestyle factors. If you smoke a lot, your odds of getting lung cancer are higher, but what determines whether you are diagnosed at stage 1 or 4? What determines how long you have after that diagnosis?
While we can’t do a similar graph for any other cancers, as 1 year survival is only published for ‘all’ and ‘the big three’, we can look at early diagnosis for the next ten most common cancers.
The same data is shown two ways. The first graph on the left shows the early diagnosis for each cancer lined up by CCG. The second shows each cancer lined up by percentage.
Looking at the first graph you can see that there isn’t much of a pattern, a CCG that has early diagnosis in one cancer has late diagnosis of another. Even if you can’t see that by eye, when you measure the correlations the surprising thing is that virtually nothing correlates with anything else. Of all of the 10 cancers shown, early diagnosis in one cancer in one place doesn’t predict early or late diagnosis in any other cancer (there are a few weak, correlations, but we can round that down to ‘nothing’). The next thing, shown more clearly in the second graph where each cancer is lined up in size order, is that while there is quite a range in the percentage of patients diagnosed early each cancer still has it’s own pattern. Melanoma is best then I’ve listed them in early diagnosis order on the graph.
Lots of things go into early diagnosis. Obviously it’s easier to spot stuff happening to your skin than to an organ buried in a bony cage, which is why melanoma is top of the list for early diagnosis and lung is near the bottom (though not in the same graph). So part of it will be your whether or not you have symptoms, then spotting those symptoms, going in to the GP, or screening to catch things before they start to produce symptoms…
Coming up to the end of the blog post, I’d like to close on the answer to all this variation, ideally with a few simple bullet points that would totally fix the problem. But unfortunately I’m all out of simple answers today (get changes in your body checked out? watch out for random bleeding? spend more money on the NHS?) It’s not that I don’t believe in simple answers, I just don’t have one here. Still if I haven’t got any good answers, hopefully it’s at least a good question.