Last week the Office of National Statistics (ONS) published the latest (2008-2012) cancer survival figures for patients in England.
Overall, these new stats were encouraging; showing that survival for many types of cancer is continuing to steadily improve.
But the story remains less positive for some types of cancer – lung, pancreatic, oesophageal cancer and brain tumours still have low survival, partly because they tend to be diagnosed at a later stage when they’re much harder to treat.
What the figures definitely did not show was signs that cancer survival is decreasing.
A closer look
After doing some digging it appears that some ambiguous wording from ONS is partly responsible for the papers getting the message so wrong.
Included in the ONS publication is a comparison of survival for patients diagnosed in the latest 5 year period, 2008-2012 and one year earlier, 2007-2011.
Clearly there’s a lot of overlap in the years that are being looked at. It’s also a very short period of time for any big changes in survival to happen, as that requires improvements in the stage at which patients are being diagnosed with cancer and how effectively it’s being treated.
It’s therefore unsurprising that the changes seen are relatively small, generally around 1-3%.
When it comes to statistics small changes need to be treated with caution – it’s possible that they could represent random variation in the figures between years rather than a genuine increase or decrease.
In the case of the decreases reported in the media (as well as some of the increases) they fail to pass the tests used to show that these changes are not just down to random variation. Put simply we can’t trust that these changes are ‘real’, or in the language of statisticians they were not ‘statistically significant’.
Is this significant?
In their publication, ONS don’t mention whether the changes they’re reporting are statistically significant or not. ONS have simply described the changes in the survival figures, which increased for the majority of the 24 cancer types that they looked at (apart from the six exceptions that were picked up by the media).
This omission means that the survival figures have been misinterpreted. We passed on our concerns to ONS but disappointingly they do not intend to add this clarity to their description of the data.
These were alarming headlines in the media, which are based on very small changes that could have been avoided with a bit more clarity and broader fact checking.
This is important to us for a number of reasons.
- If we are using statistics to examine data and using this to make decisions about what we think should be the focus of political action then these figures should be accurate and need to be interpreted correctly
- Patients could be alarmed about what this means for the care they are receiving
- We want to discourage fatalistic attitudes so patients visit their GP when they have symptoms and avoid a situation where they may believe their chances of surviving their cancer are not as good as they are.
More work to be done
The confusing way these figures have been reported has masked what should be seen as good news. Ultimately, cancer survival has steadily been increasing for many years now and these figures do not give evidence that suggests that is stopping. Since the 1970s overall cancer survival has doubled in the UK and half of all patients will now survive the disease for at least 10 years.
But perhaps most importantly, these figures clearly show that there is still a lot of work to be done – particularly for lung, pancreatic, oesophageal cancer and brain tumours. To ensure that survival does increase for all cancers – including those that are currently harder to treat – we must ensure that statistics help to indicate where the real challenges lie, while still celebrating the successes.
We’re working to accelerate progress in cancer survival for all cancers sooner. By increasing our research into these hard to treat cancers with lower survival and boosting our investment to help diagnose cancer earlier we can help ensure that improvements in survival for all cancers are reflected in the stats of the future.
Matt Wickenden is early diagnosis and communications manager at Cancer Research UK