Skip to main content

Together we are beating cancer

Donate now
  • Health & Medicine

It may seem unlikely, but could a statistician save your life?

by Matt Wickenden | Analysis

9 July 2013

4 comments 4 comments

Blackboard
These studies are complicated
Blackboard

Data can save lives

Identifying where a health service is doing well – and where it needs to do better – is one of the quickest ways to improve things for patients.

Having reliable, up-to-date figures means those responsible for running the health service can ensure that patients are getting the very best care that the public rightly demands from the NHS.

The shocking revelations about the failings of the mid-Staffordshire hospital are a painful reminder of how important this can be.

But accurate and timely information on healthcare standards can also become a vital resource for the public. Today it’s second nature to search for independent reviews of restaurants, hotels, shops etc.

So why shouldn’t the same be true for our health?

In recent years, the information available to monitor the NHS’s performance has undergone dramatic improvements – and what are called ‘cancer intelligence services’ have been leading the way.

At the National Cancer Intelligence Network (NCIN) ‘Cancer Outcomes’ conference in Brighton last month, data collectors and statisticians across the UK came together to discuss ‘cancer data’, and showed how the data revolution is beginning to put power in the hands of patients.

How do we review a health service?

So how do you know if a hospital is offering high quality care, or whether cancers are being diagnosed at the earliest stage possible?

These are the types of questions that patients, campaigners, politicians, and doctors – as well teams here at Cancer Research UK – are all asking.

In the course of being diagnosed with and treated for cancer you could see a GP, numerous doctors and nurses and undergo multiple diagnostics tests. On top of that, the cleanliness of the hospital, the practicalities of getting to your appointments and your relationships with the medics will all be important.

And this is where the statisticians are increasingly able to help by picking apart the data on all these different areas and transforming it into reliable measures of performance.

Providing the tools of the trade

In 2007, Professor Sir Mike Richards – then the Government’s National Cancer Director – laid down a challenge  after news that the UK cancer survival lagged behind a number of other countries: “Quite simply, we want to have the best cancer information service in the world by 2012.”

It was an ambitious target, but one that’s pretty much been met. At the conference Dr Jem Rashbass – National Director for Disease Registration – gave an update on the enormous progress that’s been made to develop the English National Cancer Online Registration Environment – known as ENCORE. The database, which was in its early stages development last year will allow us to ask the NHS questions that were simply not possible to answer before (there’s more in this article on the BBC website).

To protect patient confidentially, the full range of information in ENCORE is carefully controlled. But the snapshot below gives you an idea of what  will soon be possible,

The screenshot looks at data on bowel cancer patients who were referred for treatment in the East of England (the lines show where patients were moving between services). It shows the the hospital and clinics they used, and identifies other characteristics  such as the stage at which they were diagnosed

The screenshot also shows that patients are not simply treated in a single place. It also provides data that allows us distinguish the outcomes between different groups of patients. For example we can look at the effects of stage and age at diagnosis on the treatments patients receive and also look at their survival. (Update September 2013 – screenshot edited in response to reader comment below)

ENCORE-screenshot

The ENCORE system

Using this information we can identify how to improve things for patients. For example, if a lot of patients were diagnosed late then action should be taken to increase early diagnosis.

Making comparisons highlights underperformance

But as well as looking at differences within groups of patients in the same area, it can also be helpful to compare different areas, to see what’s working, and what’s not.

We’re all familiar with this on travel and review websites; comparisons offer a simple way to spot the best places to go.

One of the benefits of the improving quality and detail of information in databases like ENCORE is that it will increasingly enable us to make more comparisons between different parts of the country, hospitals and even individual surgeons.

For example, the NHS recently published information on consultant led surgical teams for a range of diseases across England including  death rates. Those  surgeons who refused to do so were ‘named and shamed’. (The information can be found on the NHS Choices website)

But this comes with a huge caveat. It’s not easy to make these kinds of comparisons in health care.

It’s not enough just to compare outcomes for patients with the same disease. Firstly – as pointed out in a recent paper in the Lancet – we need to have a large enough number of patients to know the results are a fair reflection of overall performance, otherwise the statistics are meaningless.

To compare things accurately and meaningfully, we also need to know that patients were diagnosed at the same stage and age, had similar levels of general health and other factors that influence how likely they are to survive. And we need to bear in mind that they may have attended several hospitals during their treatment.

Without these criteria the performance of hospitals or clinicians could be unfairly called into question. For example a specialist surgeon who operates on more of the most difficult to treat patients may have worse results than someone operating on patients who are easier to treat.

But these obstacles are becoming less of a problem the more data we have to tackle them.

Does more mean better?

After the whizz-bang of ENCORE, Professor Henrik Moller from Kings College London presented some fascinating statistics, looking at the differences between hospitals that treated the most oesophageal and gastric (upper GI) cancer patients, compared to those that treat the fewest. (The paper is available online)

The differences in outcomes were largest in short term. Patients who had surgery in hospitals performing the most operations were almost half as likely to die in the first 30 days after surgery than those treated in hospitals performing the least. And even a year after surgery, patients from the higher volume hospitals were still 18 per cent less likely to have died.

Although this analysis can’t explain what is causing these differences, it can show us the areas to investigate. For example:

  • Is a hospital operating on large numbers of patients better set up to care for them before and/or after surgery, and are its surgeons more technically skilled?
  • Alternatively, are hospitals with better outcomes simply more likely to have patients referred to them and therefore treat more patients, rather than better outcomes  simply being the result of treating larger numbers of patients?
  • And finally could it be that surgeons with better outcomes prefer to work in hospitals treating larger numbers of patients.

The answers to these questions may help determine how this work could benefit patients but without knowing there were differences we wouldn’t be asking them.

As this example shows, comparisons done well allow us to identify the best performers and to investigate what they are doing differently. If, by using reliable and meaningful measures, we can show a service is not achieving the best possible outcomes, then the medics themselves, those commissioning services and the rest of us can begin to answer the question: why not?

Ensuring patients are diagnosed early

But as well as making sure people in the health service get the best care, we need to make sure  cancer patients are diagnosed as early as possible –  something Cancer Research UK and others have been campaigning to achieve. And statistics can identify the groups of people who are more likely to be diagnosed late – a first step towards understanding and improving things.

At the conference, Dr David Greenberg from the National Cancer Registration Service illustrated how powerful this could be. His data showed that in England, an estimated 450 deaths from breast cancer could be prevented every year if the proportion of women diagnosed early in the most deprived groups increased to match that for the least deprived women.

This point was reinforced by Dr Paul Aylin from Imperial College London, who presented research showing that both more deprived and older patients are more likely to be diagnosed with cancer after an ‘emergency’ admission – known to be linked to lower survival rates for many types of cancer.

Again it’s possible that these groups are being diagnosed later and that some deaths could be avoided if people were aware of the early symptoms of cancer, went to see their GP when they spot them and where necessary are referred on quickly by their GP.

How can we use these stats? One way is to target awareness campaigns at these groups of people and the medics themselves, such as the  Government’s recent Be Clear on Cancer campaign, or the work we’ve been piloting around the country we discussed in this recent post.

We have the statistical power

The work presented at the NCIN conference gives us great optimism about the improvements that can be made by using data intelligently and openly.

The conference brought together people from across a wide range of disciplines within the health service – as well as from charities like Cancer Research UK – and showcased work on every aspect of patient’s interactions with health care – from diagnosis and treatment, to the quality of conversations taking place with their doctors and end of life care.

But of course the ultimate goal of all of this isn’t just data for the sake of data – it’s to provide clear and reliable measures of where the NHS is doing well and where it could do better, and ensuring we can all use that information. This is the start of the information revolution and we are beginning to make a difference.

Cancer Research UK’s own Cancer Statistics section of our website is the leader in UK cancer statistics and we’ll be making sure we help make all this new information available to everyone. Yesterday we launched an exciting new addition to our website which makes local statistics available to health care commissioners, politicians and everyone else interested in them.

We statisticians are doing much more than providing a market for “I love spreadsheets” mugs: we’re using patient data to save people’s lives.

Matt

  • Matt Wickenden is a senior statistical officer at Cancer Research UK

Image of ENCORE provided by Dr Jem Rashbass; other image via Wikimedia Commons

    Comments

  • Matt Wickenden
    9 September 2013

    ro armstrong: thank you for notifying us of this issue and apologies for the delay in responding to you. After consulting with the team responsible for ENCORE and that supplied the screenshot we can confirm that the heading for that data should be “Weeks from urgent referral by a GP/dentist to cancer diagnosis” (as shown in the corrected screenshot). The analysis system is still under development, which is why this error was present.

    If you’re interested in finding out more about the time patients are waiting for treatment, we provide a range of information on this for specific areas on our local cancer statistics website under the early diagnosis section. Quick access to the best treatment can improve outcomes for patients and improvement is needed across the UK.

    Matt Wickenden, Cancer Research UK

  • ro armstrong
    9 August 2013

    Something wrong I think with your data on ‘weeks to first treatment’ unless first treatment is defined ‘diagnosis’. Cancer sufferers known to me personally (four this year, 1xkidney, 1xbowel, 1xovarian and 1x oesophageal) have waited at least six weeks from diagnosis to first treatment.

  • Zinia Thomas
    10 July 2013

    Thanks Matt for sharing your views. Here you illustrate some of the awesome techniques of health review. many people will be benefited. Thanks again.

  • Celia
    9 July 2013

    I think it’s time we were all told the truth about the causes of cancer. It’s now affecting nearly one in two people. In 1970 it was one in thirty. We need a cure not figures about how many people are being cured. This is misleading as far more people are getting it. And the drugs companies are making a great deal of money from treatments.

    Comments

  • Matt Wickenden
    9 September 2013

    ro armstrong: thank you for notifying us of this issue and apologies for the delay in responding to you. After consulting with the team responsible for ENCORE and that supplied the screenshot we can confirm that the heading for that data should be “Weeks from urgent referral by a GP/dentist to cancer diagnosis” (as shown in the corrected screenshot). The analysis system is still under development, which is why this error was present.

    If you’re interested in finding out more about the time patients are waiting for treatment, we provide a range of information on this for specific areas on our local cancer statistics website under the early diagnosis section. Quick access to the best treatment can improve outcomes for patients and improvement is needed across the UK.

    Matt Wickenden, Cancer Research UK

  • ro armstrong
    9 August 2013

    Something wrong I think with your data on ‘weeks to first treatment’ unless first treatment is defined ‘diagnosis’. Cancer sufferers known to me personally (four this year, 1xkidney, 1xbowel, 1xovarian and 1x oesophageal) have waited at least six weeks from diagnosis to first treatment.

  • Zinia Thomas
    10 July 2013

    Thanks Matt for sharing your views. Here you illustrate some of the awesome techniques of health review. many people will be benefited. Thanks again.

  • Celia
    9 July 2013

    I think it’s time we were all told the truth about the causes of cancer. It’s now affecting nearly one in two people. In 1970 it was one in thirty. We need a cure not figures about how many people are being cured. This is misleading as far more people are getting it. And the drugs companies are making a great deal of money from treatments.