Follow the Science, Follow the Data

Earlier in the year, a casual conversation with a colleague in Croatia uncovered a controversial viewpoint. He said that many of his patients and members of the public were more afraid of a COVID-19 diagnosis than a cancer diagnosis. I’d heard similar rumblings from colleagues in the UK, but there were no data to say whether this was true or not. So, we set about collecting data that would give us a precise insight into the effect of the COVID-19 pandemic on cancer services and cancer patients in the UK.

There seems to be a misconception that data collection always needs to be complex, burdensome and requiring months of in-depth analysis to yield anything truly useful. With a global pandemic rapidly growing, we needed to do something straightforward and simple to give us the immediate answers we needed to make informed decisions that would ultimately help save lives.

We chose two parameters that we could measure on a weekly basis across the entirety of the UK, comparing our results with pre COVID-19 data so as to capture the direct effects of COVID-19 on both the diagnostic pathway and the treatment pathway of cancer. Collecting data in real time would reveal the effects of the pandemic as they were happening. The two parameters we chose were:

• The ‘2 Week Waiting’ or ‘red flag’ referral time for cancer (our early warning system to confirm or dismiss the suspicion of cancer)
• Chemotherapy attendance (our proxy measure of the ‘health’ of the cancer treatment service).

Shockingly, we saw that cancer referrals had dropped by over 70 per cent. I realise there’s a danger when we read statistics so often that they don’t feel personal. To be clear, this means that seven out of 10 people who potentially had cancer were not being referred on to confirm if that suspicion of cancer was correct or not.

Looking at chemotherapy attendance, again, we saw a significant drop of over 40 per cent. In other words, four out of 10 people who were meant to be receiving chemotherapy at a particular time were not getting that treatment, due to delays caused by the pandemic.

The advantage of looking at real-time (or ‘near’ real-time) data is that you can continue to monitor the situation as it is unfolding and use that information to shape the response for early intervention. If your data is 3 months old (or more) you’re really trying to fight the effects of COVID-19 on cancer with one hand behind your back. We’ve been able to closely track the effect of COVID-19 on cancer services as we went into lock-down, as we have slowly emerged from lock-down and as cancer services have begun to be restarted (based at least partially on our data above).

We’ve seen a positive recovery of cancer referrals and chemo appointments, but we’re still not back to pre-COVID-19 levels. The longer the figures take to get back up, the worse the situation is, both for the citizen waiting for a diagnosis and for the cancer patient waiting for treatment. The later that cancer is picked up, the worse the outcomes, in terms of both excess deaths and other underlying health problems or complications associated with more intensive treatment.

What we’ve also been able to do now is create a model based on our real-time data and run this model in a data set from four million patients from England, a proportion of whom are living with one of 24 different cancers, to predict what happens 12 months from now in relation to individuals with cancer. We apply different data-informed scenarios – best case, worst case, most likely, least likely – tested on real data, not a hypothetical best guess of how COIVID-19 is affecting our services. This way we are able to predict the overall excess risk of dying for people with cancer as a result of the COVID-19 pandemic. Our data suggest that we are likely to see between just over 7,000 to just under 18,000 excess deaths in the most likely scenario, but this could rise to as many as 35,000 in the worst-case situation.

Worryingly, we are facing significant issues down the road as well as at this present moment, but the power of real-time data allows us to develop informed solutions, which will help save lives in the long-term. For example, data is helping us in our work in colorectal cancer, the second biggest cancer killer in Europe, where we can use the intelligence that we are collecting to decide which patients we should prioritise for colonoscopy. This is really important as the backlog that we currently face cannot be adequately addressed, even if we returned to pre COVID-19 colonoscopy rates. So, we need to select those citizens who are at highest risk, to maximise the use of our screening/diagnostic services. Data show us where we need to prioritise our resources and where we have greater flexibility. There is no denying it: data saves lives.

So, if something good is to come out of COVID-19, it should be an acceptance that a mixed approach to data collection and analysis makes sense. There is always going to be a need for the most robust accurate, ‘clean’, datasets collected and analysed over significant periods of time – but we also need to embrace rapidly-accessed, real-time data that shows us what is happening in the here and now so that we can use it to make rapid informed decisions that positively influence patient care. So I exhort you to Follow the Science; Follow the Data. Don’t make me fight COVID-19 and cancer with one arm behind my back.

This blog was originally published by Data Saves Lives and they have kindly given DATA-CAN permission to reproduce it.