Collaboration for Academic Primary Care (APEx) Blog
Collaboration for Academic Primary Care (APEx) Blog
Posted by ma403
12 February 2025They say a tree falling in a forest, but with no-one to hear it, makes no sound. Well maybe, though I’ve always doubted it. What’s clear is that medical research, if it doesn’t change practice, makes no sound. In this spirit was a conference in London last month.
I co-led a successful HTA research bid three years, with Sarah Bailey and Gary Abel as co-applicants, alongside a host of other colorectal cancer luminaries. This grant was to see if we can improve the prediction power of colorectal cancer in patients whose symptoms suggest cancer is a possibility. Currently, there is a test called faecal immunochemical testing (or FIT, which allows all sorts of duff puns, that I’ll avoid). This FIT test identifies haemoglobin in the stool, the idea being that colorectal cancers leak blood into the stool. The test works – as the Exeter team (and others) have shown. But can it be improved? At present, there’s a single cutoff, 10mg haemoglobin/g faeces, to define a ‘positive’ test – and thus to trigger a colonoscopy. Age isn’t considered, gender isn’t considered, other symptoms aren’t considered…..
So, we wanted to see if other variables improved the prediction beyond the simple use of FIT alone. We did a systematic review https://doi.org/10.1016/j.eclinm.2023.102204 which showed lots of people had done small studies, though none of the algorithms/equations had been successfully validated in another population.
We then created a new algorithm from a huge Nottingham dataset, having tried in a smaller Exeter/Sheffield dataset. It included age, sex, FIT, and two things from a simple blood test – platelets (an Exeter speciality https://www.mdpi.com/2072-6694/16/17/3074), and MCV (if you know what that is, take a bow and a chocolate biscuit). It seemed to work https://doi.org/10.1111/apt.18459. We then tested it in three other datasets (a later Nottingham one, an Oxford one, and one from NW England). IT WORKED!
We did a health-economic study (almost accepted) and two patient/practitioner qualitative studies (the latter also almost accepted, asking the question ‘are patients happy with an algorithm dictating whether they proceed to colonoscopy: answer – maybe!).
All sounds good: now to tell the world. We invited ‘the great and the good’ particularly targeting NHSE (as they have the NHS chequebook, so to speak), but with a good number of patient reps (to keep us on our toes – thank you!) and academics (to ask us difficult questions – what’s new?).
IT WORKED! I say this because we got genuine dialogue with the whole audience. Of course, we’d kept NHSE and NICE etc., in the loop (tip: do this as much and as often as you can). There were remaining problems…could we be sure the apparent 17% reduction in colonoscopies but no fewer cancers found would actually happen if we went live? Would we miss other conditions, like inflammatory bowel disease? Who would get MHRA approval? [ this last is a really tiresome
process, originating from some wiseacre deciding medical equations were medical devices – like a pacemaker, or artificial hip so needed to jump a million hoops for approval].
At the end of the day, there was a feeling of ‘we’ve done our best as researchers – now it’s over to policymakers,’ but we were actually delighted to be green-lighted by NHSE (and to get off the hook of our having to get MHRA approval).
The message? End of programme events are good. As ever, they require preparation, but the end result should be better colorectal cancer services for all, in that the 17% reduction in scopes means the 83% get it more quickly, the 17% don’t have the worry of a colonoscopy (or the ghastly preparation for colonoscopy), the colonoscopists actually get time to draw breath between cases, and taxpayers save a little bit of NHS dosh.