Proton Therapy Congress 2016
How do cancer registries operate in the UK?
Dr Jem Rashbass – Public Health England, UK
Cancer registries have been around for a very long period of time, in England they’ve been around for 50-60 years and they were initially set up to monitor the safety of the population for radon exposure. What they actually are is the most phenomenal resource of cancer data collection and over the last ten, twelve years I’ve run a modernisation programme for the cancer registries in England so that we now have, rather than just a cancer registry, a cancer registration service which collects data on every patient in the population, nearly half a million cancers a year, across the population of England and it’s no longer just a thin dataset of a few data items but what we are looking at is the treatment, the outcome, the risk factors for every cancer patient in the country.
This sounds like a lot of data
It’s quite a lot of data, it depends how you count your data. It’s not large by terabytes size, if you want to collect x-ray images, which we don’t collect, you can have a lot of data but it is a lot of data items and it’s also a lot of complex data so we have up to, well, probably over a thousand data items on every tumour. We are interested in tumours rather than just cancer patients, because obviously over their lifetime a patient may have more than one tumour. What we do is we work at the level of a tumour and we then understand the treatment associated with that tumour. Once you’ve got all of the details about the patient themselves, their treatment, their chemotherapy, their radiotherapy, their medical history, you’re looking at a thousand data items on each tumour. Cancer registries are not a research project, what they are is a national resource for protecting the public’s health and a resource for improving healthcare and healthcare outcome. I think they have come into their own in the last five, six years as we move towards personalised medicine. There is a certain irony in that personalised medicine is obviously about an individual and here am I collecting data on a whole population. The challenge is that for a whole population in order to understand each individual cancer you now need to know about every cancer because cancer’s becoming a rare disease for individual people and that’s what personalised medicine is about. When I see a woman in clinic with a breast cancer, I need to understand how that woman’s breast cancer is similar perhaps to the dozen or so other women in the population who might have a similar cancer and that’s why you need this very, very large population data set.
What are the next steps?
We focussed on the data we know as it were; what your diagnosis is, what your treatment is. The challenge is obviously outcome and traditionally we’ve used a rather crude measure of outcome which is just survival, but actually what we want to be able to know is what is the quality of life between this diagnosis and death; what’s your mean disease free survival? How long before you relapse? What are the complications? We’re beginning to get data to fill that in. Once we have that on the actual population we can then use machine learning algorithms to take the patient who is sitting in clinic and rather than just saying, ‘You’ve got this type of cancer, at this stage, we expect you to live this long,’, ‘Here is a profile of what we think your life will look like over the next five, ten, fifteen, maybe even twenty years, based on patients like you and the information we have collected over time’.
Have there been any issues with the data?
The challenge is around the issues of confidentiality, patient consent, there is a very significant and appropriate privacy lobby that patients need to be confident that their data is not going to be abused, that their confidentiality is not going to be breached and that is my responsibility. As a doctor my absolute responsibility is to protect a patient’s confidentiality as if they were the same person sitting in front of me in clinic. That duty applies to me and all my staff in the data that we collect. We are very clear that we will not release anyone’s data in a format that could identify them unless there are very strict permissions for doing that, and one is their consent to do that. The other is if there is a legal gateway so we can be instructed as a legal method to release data, but in general what we will do is we will release de-identified data so that we are sure that someone’s identity cannot be revealed from their data set and that’s the absolute responsibility. We also don’t use it to make money. The government policy is we do not sell data for profit.
What are your thoughts on the conference?
When you have a new area of emerging treatment, one of the things you really want to have is good data to be able to understand the outcomes and interventions that you are doing, and so I think what we discussed at the conference and the feedback is around how do you use the sort of data collection systems we’ve got to inform futures for proton therapy?
Any final thoughts?
To me what I am interested in is producing a resource that sits on the back wall of the clinic so that the doctor somewhere in the world, sitting with the patient in front of them, is no longer relying on their previous medical knowledge and focal trials to say, ‘Our best guess is that we should treat you like this, although you don’t really fit into the trial’. It is actually, ‘On all of the evidence of every patient we have looked at across the population, using data that is most up-to-date, I can tell you that the best way to treat you now is like this, and the most likely outcomes are the following’, and I think that is what I want to achieve.
