A recent HTN Now webinar took an in-depth look at how technology is transforming skin cancer care at the Royal Cornwall Hospitals NHS Trust, exploring successes in reducing two-week wait referral time from 57 days to 12 days, looking at challenges for implementing this approach, and sharing advice to other healthcare organisations looking to implement similar digital pathways.
Sandy Anderson, consultant dermatologist at Royal Cornwall Hospitals Trust, and Gabi Cohen, director of delivery at Isla Health, joined us to share their insight and experience on the project, which won in the HTN Awards category Case Study of the Year, before moving on to take questions from our live audience.
“I’m a locum consultant at Truro, where I’ve been for about 15 years,” Sandy told us, “but before that my background was in general surgery, then general practice, and then dermatology, becoming a consultant through the Portfolio pathway.” His main passion and experience has been in skin cancer diagnosis and complex surgery, he added.
Gabi also introduced herself and her role, stating: “I’m the director of delivery at Isla, and the delivery team is responsible for implementation and evaluation, focused on all of the things that sit behind brilliant projects like these. Our team make sure that at every step of the process, you have that really rigorous implementation and strong clinical leadership, working collaboratively to make things as easy as possible for organisations like Royal Cornwall.”
Innovating skin cancer pathways at Royal Cornwall Hospitals
Sandy then began his presentation on the community lesion imaging clinic initiative, a project he says “began over a year ago” as a means of overcoming the challenge of “escalating skin cancer referral rates and a shrinking workforce”. Whilst 25 percent of consultant dermatology posts in England are unfilled, he shared, the situation is “particularly difficult” in Cornwall due to its remote nature, “and in the last ten years, our skin cancer two-week wait referrals have gone from around 3,500 to 10,00o per year”.
Up until this initiative was introduced, a traditional model has been in use, Sandy went on, “where we didn’t particularly screen two-week wait or suspected cancer referrals – we just saw them all face-to-face”. A key challenge for Cornwall is also that it has “no plastic surgeons”, and less than two whole-time-equivalent substantive consultants, he shared, “for a population of about half a million”.
The model that the trust looked into was a “low cost, high volume, teletriage” approach to suspected skin cancer, according to Sandy, “and from the start we had to make it more efficient for doctor time, it had to work with what we already had in the absence of lots of funding, and it had to be acceptable to patients.”
With the intention of capturing as many of the trust’s two-week wait patients as possible, inclusion was set to cover all adults, with a maximum of 3 lesions noted by the GP to avoid “the risk of imaging the wrong thing”, and excluding genital lesions. Highlighting results from January to May 2024 and the initiative’s first 1,398 patients, Sandy told us how 24.5 percent had been discharged to a GP with advice, 38.9 percent had been added to the day case list for biopsy or excision, 34 percent had been triaged to face-to-face review, 1.3 percent had a follow-up arranged, and 1.4 percent had been referred to a surgical specialty.
“From seeing 100 percent of patients face-to-face, we reduced that to avoid 65 percent of those initial face-to-face reviews,” Sandy shared, “which freed our team up to do the rest of dermatology”.
Sandy highlighted the findings and impact at Royal Cornwall Hospitals, saying: “We think it’s a safe and effective way to operate; it’s reduced our face-to-face appointments by two-thirds, missed cancer rates are similar to our face-to-face work, and as a by-product, we’re building up a big collection of high-quality dermascopic images and metadata which will be ready to contribute to nationwide databases for testing and training AI.”
Implementing the pathway: Things to consider
“We work with a number of teams when implementing this,” said Gabi, “from contracting through to information governance, clinical safety, and to mapping the interoperability required to do this project well – in this context it’s a bidirectional HL7 feed – through to clinical mobilisation and getting everyone trained and confident with the platform, and then into evaluation and business as usual.”
The key is understanding what it is that you’re trying to impact or change, Gabi continued, “as well as the mechanisms that sit behind that, so we can look at the suite of functionality Isla can offer and make sure we’re offering the best of breed pathway”.
“For us, the bottleneck was around the availability of specialist dermatologists,” Sandy considered, “and it’s very difficult to recruit those, whereas it’s quite easy to train nurses to do biopsies and excisions, and the Isla software allows us to mark-up images electronically and complete our standard booking form on the normal hospital system, so when they come back to these day case lists, they’ve got a marked image and that booking form.” Standard surgical questions are also included in the CLIC questionnaire, he went on, “so that’s all present there, too”.
One of the motivations for the initiative for Sandy was “making it as quick and easy for the doctor as possible”, he shared. “If we made this time-consuming and laborious, it wouldn’t be any more efficient; Isla’s been really good for allowing us to see large numbers of patients quickly, and part of the reason is we’ve worked with the team to adapt the product for our needs.” It now takes less than an hour to triage, list and “do all the actions, including letters and things” for a four-hour clinic, he stated, “and the HCAs see 15 patients in clinic, which is a fair sized lesion clinic, and it takes me about 45 minutes to action that on Isla”.
Gabi shared some insight into the impacts around efficiency from a tech perspective, pointing to the importance of being able to “meet patients where they are,” especially in cases with remote populations like Cornwall. “You’re preventing them having to travel for appointments, and inherently there’s a tech need for that, so data can be accessed really securely and safely from these remote locations,” she said, “and having that really clear need made it obvious why this needed to be a digital-driven implementation.”
Offline access has been released this week by Isla, Gabi shared, “and that’s incredible exciting; not just for dermatology, but a lot of our community nursing teams work in rural parts of the country, where there might be no access to the internet”. That allows images to be stored until connectivity is available, before being uploaded onto the platform, she said, “and we don’t require patients to download anything, so the intention there is to make the burden on the patient as minimal as possible.”
Making sure that there is no added clinical burden or overhead has been an important consideration for Isla, Gabi told us, “and another key factor has been in making sure that everyone on the pathway is clear on what their role is, because as Sandy said, we’re leveraging lots of different members, and training is the priority to ensure everyone knows what their pathway looks like, and their place within that.” For this iteration, training was completed with Sandy and other consultants, and videos and user guides were used to create a module for the HCAs, she said.
A lot of work was done on making the pathway as efficient as possible, Sandy shared, “because when we started, we were uploading referral letters by hand into Isla, as I felt the triaging doctor should just be within Isla, and shouldn’t have to have four different systems open, so as much as possible has been kept within Isla, except booking the procedures, although even that is now automated to pull referral letters straight from ERS into the patient’s Isla record when a patient is added to our clinic.”
From a clinician’s perspective, Sandy continued, “within Isla I can have the referral letter there, the questionnaire data, the images, and I fill in a quick review form detailing my opinion and outcomes, and then we just select everything and output it as a PDF back to our EPR.” That has the added benefit of meaning that when admin staff are “outcoming the clinics and doing letters,” they don’t necessarily need to be on Isla, they can choose to stay in the EPR, he added.
Measuring success
Answering some incoming questions from our audience around measuring success, Sandy told us how the team is currently allowing six minutes on average for triaging a single case, with job plans accounting for “90 minutes to triage a four-hour clinic”, which also covers the admin time for processing results.
“When we’re looking to capture the impact of any implementation, we always start off thinking about safety and accessibility,” Gabi considered, “and then we think about clinical productivity and any cash implications”. For this two-week wait pathway, “there are some very specific KPIs like referral to treatment,” she went on, “as well as other efficiencies such as how long it took to action the patients.”
Keeping patients and patient experience at the centre of the model is integral, Gabi said, “and we’re making sure we’re listening to feedback and iterating on the model to make sure we’re really thinking about what that patient experience looks like”. Alongside that, it’s considering the clinical experience and usability, with co-production with clinicians a regular feature and discussions around tweaks that could be made to make the tech “more effective, more efficient” on delivering key outcomes.
Sandy shared some feedback from “a snapshot of between 100 and 200 patients” given feedback questionnaires, acknowledging “coming to the hospital and parking seems to be the biggest barrier to accessing healthcare.” Estimating that it was “about 96 percent positive”, he said: “A lot of the feedback was about being able to be seen locally, feeling listened to, and having everything moved along quite quickly.”
Ensuring successful implementation and adoption
Having rigorous governance in place when approaching this kind of initiative and assigning clear roles and responsibilities is essential, Gabi noted, “as well as having input from admin and service managers on what standard operating procedures should look like”. It’s normal to “have some anxiety” around a new process, she said, “so we had a regular weekly forum where we discussed any challenges and worked through those in a really methodical way – key to that is having the right people in the room.”
Building acceptance and encouraging adoption is helped along also by ensuring alignment with wider organisational priorities and strategies, Gabi shared, “so we’re articulating the impact something is going to have on patient populations, getting that weekly governance in place, and making sure we can address challenges quickly and effectively.”
“I have found Isla has been very responsive to any challenges we might have,” Sandy agreed, “including in the beginning where we had occasional bumps and glitches – they’ve always been fixed within a day or two, and I think that relationship has been key to the implementation.” Running “dummy clinics” first and testing out things like admin processes and image capture was important, he said, “and we’ve grown it from there, starting with one clinic a week and then adding clinics at different sites.”
Everything that is put into Isla is coded, Sandy told us, “it’s all drop-downs, so there’s no free text, and I can get Excel spreadsheets and CSV files at short notice, either by myself, or via request if it involves big numbers.” That functionality has been “brilliant”, he went on, “and we’ve been able to present and publish a few things so far using that, which is very different to other software I’ve used, and far more reliable than any of the hospital outcome data I’ve encountered before.”
Key takeaways and advice for those considering similar implementations
Moving on to consider what advice they would offer to anyone on the call looking to implement a similar pathway, Sandy said: “There’s no need to reinvent the wheel – get in touch, you can share our processes, share our training packages.” It’s been “a very open and transparent process”, he went on, “and I’ve seen all the costs and the IT security side of it, the data ownership, and so on, so you just need to keep things transparent and honest.”
Planning is key to success, Gabi considered, “and also from our side, it’s having flexibility to tweak things or update them, being open to iterating as we work through the project, learn new things, and review data.”
When considering safety and risk when implementing a pathway like this, Sandy said: “The HCAs have got clear guidance on that if there’s any ambiguity or uncertainty, the procedure should be abandoned and a face-to-face review should be arranged. Similarly, if a patient tells them they have another 12 moles on their back that they would like checking by a doctor, the procedure is abandoned.”
Discussing how the Isla platform can support data security, Gabi pointed to Isla’s focus on ensuring data can be captured and stored in a “really secure way”, before being made ‘immediately accessible to the different teams that need access to that data”. As a cloud-based platform, she said, “the upload process is encrypted, and nothing is stored locally on any device – when using iPads, for example, that information is uploaded straight into the Isla solution”.
The panel also looked at the potential for the images collected and uploaded to Isla to be used for research, with Sandy noting “lots of opportunity” for AI in dermatology and skin cancer diagnosis. “From the start, we’ve asked patients to consent to their images being uploaded to their medical record, but then offered a separate consent for those images to be anonymised and used for teaching and training, including in the training of AI.”
Having collected between 10,000 – 20,000 dermoscopic images of lesions, “we’re now going back and adding their diagnostic tags”, he went on, “and my vision, along with others, is that we end up contributing to an anonymised nationwide research database that we can use to test and train AI products on real world data.”
One key takeaway from an implementation perspective is that “there isn’t one solution that fits every single patient population or every single trust we work with”, Gabi considered, “so getting an idea of how this might look in your organisation and mapping out the problem you’re trying to solve is really important”.
“My key takeaway would be not to be afraid of this being clinician-led,” Sandy said, “because as much as this is about technology, this pathway will be successful if it’s driven by clinicians.”
