Polderen Part 2 - more on PROMS
- Allan Wardhaugh
- May 31, 2024
- 6 min read
This was the follow-up piece to the Polderen Blog published for the same group (CiMAT) during my time as CCIO in Cardiff and Vale UHB. CiMAT was a stakeholder engagement group from which leaders were sought and devloped. It had around 300 members, with some guests from other health boards.
I want to follow up last week’s blog* continuing with this theme of patient reported outcome measures, or PROMs.
The ICHOM conference provided an opportunity to hear international examples of how PROMs are being used in practice to give new insights into services and offer opportunities to reshape them and derive more value for patients. I’ve highlighted three presentations below. A disclaimer is that these are taken from my rough notes on the day, and I have not yet been sent a link to slides so I can fact check – but the messages are general rather than specific.
Uncovering the hidden agenda
First up, an interesting presentation from Shehzad Niazi a psychiatrist practising at the Mayo Clinic in Florida. Before he even started to speak specifically about PROMs, he spoke a little bit about the use of the electronic health record. In his case the use the Epic EMR, but his comments were more generic. His aside was that surveys demonstrate that physician time in the US is now approximately 75% spent populating the EH are and doing other desk work, with only 25% of face-to-face clinical time. It was interesting to note, although it’s not the first time I’ve picked up on it, that the EHR often has quite negative associations for doctors in the US (see Robert Wachter’s book ‘The Digital Doctor’). I couldn’t help contrasting this with the clamour among many clinicians I interact with to have our fully functioning EHR, and of people like me to advocate for it. As usual, as my granny and (everybody else’s granny) used to say-be careful what you wish for. I suppose the point is the HER can simply be a time-consuming obstacle, just like ‘paperwork’, and is only beneficial once the information it harnesses is used in ways that allow real benefits.
The main part of his presentation was actually about how collection of PROMs have been facilitated by using an electronic form to collect information from patients, although his team’s journey had started with pen and paper. The advent of electronic collection improved the return rates very markedly and have enabled the collection of significantly more patient outcome data than was previously possible. Example was given of their use in a liver transplant service. In the world of liver transplant follow-up, it is recognised that the presence of depression in patients increases mortality. They used a PROM which mirrored an ICHOM standardised PROM for depression and anxiety, and the information was captured at predefined time points in the patient pathway and the results presented to the clinician. This demonstrated that there were very high prevalence rates of depression among the patients, and the use of PROMs made it more likely that the issues would be discussed at the follow-up visits with their clinician, giving an opportunity for appropriate exploration and if necessary, interventions. The key message was that by using PROMs data in real-time to inform individual patient consultations, new information was exposed that would otherwise have been hidden and not discussed. This reflected previous research demonstrating that mental health co-morbidity occurs in around 70% of patients with cancer, but that clinicians are poor at picking these problems up during routine clinic visits without the aid of patient outcome measures.
I was reminded of early training I had in my GP years on the importance of uncovering the ‘hidden agenda’ in a consultation – finding out what the patient really wanted to talk about, rather then the declared reason for the appointment. Of course, what you do, and whether you can intervene having discovered this agenda is another story. The suggestion from this presentation was that in a specific group of patients it was proving beneficial – and it is the transition from paper to digital that has provided the step change.
A postscript was that their successful implementation of electronic PROMs was now being used in other clinical areas but was starting to come up the against an obstacle. The electronic system used was a stand-alone one (if I understood correctly), and not integrated with the Epic EMR! This is why, in Wales, we are so keen to ensure that electronic PROMs are integrated with our emerging, modular EPR. It’s encouraging to note that PROMs collection is enabled by patient portal solutions available to us, which I’ve mentioned previously (Patient Knows Best, Dr Doctor and many others), and that the national PROMs team in Wales are working with partners in the Digital Health Ecosystem Wales to achieve such integration.
Over-coming questionnaire fatigue
One potential obstacle to collecting information about patient outcomes, and particularly where wider health measures might also be recorded is that of ever-lengthening questionnaires. I’ve encountered this myself in meetings and conferences that I’ve attended where it sometimes feels like I need to have a special dedicated session simply to fill in the feedback for the conference organisers. Although a real issue, there were lots of stories at ICHOM describing how using PROMs to inform every patient consultation improves patient compliance. Patients quickly recognise that their feedback is of value, taken seriously, and leads to meaningful discussion and decisions around their clinical care. They will therefore continue to make the effort to complete the questionnaires that record PROMS.
Nevertheless, there is always room to improve so I was also interested to hear a presentation from a doctor working at Brigham and Women’s Hospital on a product called (I think) ImPROVE [do you see what I did there?]. He had developed a PROMs tool for use in breast cancer clinics, specifically to inform discussions about breast reconstructions following mastectomy. He was clearly very talented in the field of informatics and coding – smarter than the average Clacker - and had developed a product following careful user-design principles and used machine learning algorithm to produce our questionnaire which swiftly and seamlessly adapted the questions asked according to information being provided. In effect it was reproducing the conversation a patient might have with a ‘wise experienced clinician’, who asks the right questions and only the right questions with nothing superfluous. The purpose of the questionnaire was to try and explore issues important to the patient and improve post-operative satisfaction with results, and the model certainly seems to be achieving this from the data presented.
Another particularly interesting point this presenter made related to a similar study that had been done in the Netherlands using a large and very rich clinical dataset, but, only clinical. This dataset did not contain any PROMs type data, and despite the comprehensive clinical data the predictive value was very poor. The message was that without the patient reported data the value of clinical information is limited. It’s what the patient reports that matters.
Whether the PROMs datasets developed by ICHOM can be used as the basis of machine learning and have predictions based upon them is still speculation, but it’s not difficult to see that this could be an interesting area of research and learning in the near future.
Harnessing data
Moving away from PROMs and back to clinical outcome data, this time on what can be achieved by using large pools of data at national level when it is collected electronically. The keynote speaker on was Prof Steffan James from Uppsala in Sweden. Those of you in cardiology will be aware of this study (TASTE, NEJM 2014) but it was new to me. He described how the combination of disease registries across Sweden allowed the pooling of clinical outcome data for a large number of patients undergoing an expensive technical treatment of thrombus aspiration from the coronary artery after presenting with ST elevation myocardial infarction (STEMI). At the time this treatment was becoming widespread in Sweden and was felt by many to be very effective, although it cost more than conventional treatment. By looking at data that had been ‘harmonised’ (i.e. standardised) it was possible to look at the outcomes for patients across Sweden. The disease registries were populated electronically, by what was in effect an EPR, a simple element was introduced to the EPR form where the clinician could ask the patient as part of the initial assessment if they wished to participate in a randomised controlled trial, and within seconds of receiving consent the EPR would generate the randomisation allocation, and then facilitate the recording of outcomes using the national disease registry. This enabled the running of a very large randomised control trial across Sweden in a very rapid time for an RCT, at much reduced cost. The results showed that the expensive treatment was no more effective than the standard, enabling the resource being directed at this treatment to be reallocated to somewhere else of greater value.
The accompanying editorial describes this technique of the randomised registry trial as the next disruptive technology in clinical research. I recommend reading it, as it describes better than I can, how this ability could revolutionise RCTs. The potential that this could become possible in Wales in the very near future if we can successfully establish a National Data Resource, combined with a good user-interface for entering data, which will be an order of magnitude greater than at a disease-specific registry.
The two essential components seem to be standardised data collected from multiple centres, and the ability to enter that data live while undertaking the clinical interaction. Building that standardised resource may take time, resource, and may be a bit un-sexy and invisible, but it’s going to provide the foundation to everything we can do digitally in the future, so it’s worth getting it right.
Comments