Published on the 19/10/2023 | Written by Heather Wright

Public health and workforce benefits, while reducing inequality…

Kiwi primary health organisation ProCare is trialling robotic process automation ‘digital assistants’ to help take the burden off overworked GPs – but Paul Roseman, ProCare general manager strategic development is very clear: It’s not a tech initiative.

“You have to look past RPA being a technology and think about it as a workforce initiative that promotes workforce satisfaction and retention and promotes improved clinical outcomes including reducing inequalities,” Roseman told iStart.

“It’s time returned to the practice to see patients and deliver other care.”

The project, which took ‘a smell of an oily rag, New Zealand number eight wire approach’, has been clinician driven. Don’t go looking for large tech teams or outsourcing to service providers here (more on that later!).

ProCare is New Zealand’s largest primary health organisation (PHO) with around 174 practices and 700 GPs in its network, servicing 850,000 enrolled patients. 

Over 18 months the PHO has developed three archetype bots, two of which it has progressed development with and the third where Roseman says it has just ‘played’. Ironically, the one bot which you might most typically associate with RPA – one capable of reconciling payments – is the one ProCare has ‘just played with’. 

“It doesn’t generate clinician time, because clinicians are not the ones doing reconciliation,” Roseman says. 

It’s garnering time back for clinicians that is the key focus of ProCare’s RPA work. 

The project kicked off after clinical direct Jamie Shepherd, an Auckland GP, observed that the volume of emails hitting his inbox had swelled by more than 80 percent between 2017 and 2022. At the same time GP feedback noted the amount of time they were spending doing after hours work to stay on top of admin and messaging associated with patient care, creating burnout, and a lack of workforce to fill all the roles.

Dr Shepherd – who is ‘tech-interested’ – identified a number of areas where RPA could work for the PHO, and he and another staff member started to experiment with freeware versions of the technology.

Deciding there was merit in trialing the technology, a more formal evaluation began with UI Path eventually chosen.

Enter the bots

While the reconciliation bot might not be a big winner for ProCare, the other two – one handling emails and one performing cardiovascular disease risk assessments – are proving their worth.

The inbox clearance bot identifies messages that don’t require any clinician action or decision making – a normal result on a routine breast screen examination for example, where the clinician wasn’t looking for something, but simply doing routine screening. 

The bot can find the result, read it, file it, notify the patient of the normal result using a standard practice process, and set a recall for the next exam.

“By doing that we take out an estimated two minutes of nursing time for each result and that’s returned to the practice to see patients and deliver other care,” Roseman says. 

While that two minutes might not sound much, it quickly adds up Roseman says, noting the potential benefits during any future outbreaks, such as a measles epidemic. 

It’s the third set of robots, however, which hold the greatest time savings and greatest population health benefits.

The cardiovascular risk screening bot makes around 120 decisions each time it runs, logging into the patient management system (PMS) and calculating risk scores for patients in a process that is saving five to six minutes of clinician time per record – and contributes to how clinicians spend future time.

“The point of a cardiovascular risk screening is to identify people at risk of heart disease, so if you spend five to six minutes with 20 people to find the one person you need to work with, you have effectively wasted more than an hour that you could have spent with that person.”

Roseman says some of the practices under pressure and trialling the bot have been more than doubling their screening rates.

As well as enabling patients with potential cardiovascular risk to be picked up earlier and provided with more in-depth assistance, the bot has another valuable role: The practices under pressure – which are seeing the greatest benefits – are typically those looking after the poorest communities, and those communities usually have the poorest health.

“Technology is often accused or risking exacerbating inequality by being unequally available, and I would argue that this is an excellent example of technology being able to improve equality,” Roseman says. 

Fourteen practices have been running at least one of the bots. Roseman says ProCare is now assessing whether it has a sustainable and viable model for bots. 

“Assuming we make that decision and do proceed, there is a lot of potential growth in developing the existing robots’ capabilities and they way they work, plus growth in new robots and also there is growth in application to other platforms.”

The current robots only work on one PMS system – the native browser based Indici system – and Roseman is keen to see that extended. 

He’s also keen to see other PHOs embracing the technology in collaboration with ProCare.

“The issues we have around workforce and capacity in healthcare is not just a ProCare issue and we’re not here to solve this just for ProCare, we want to contribute something. Involvement of others in what we are doing will enhance the quality and efficiency with which we are able to do that.”

The robots have been developed without the assistance of any outside service providers, and while ProCare has an internal tech team, Roseman says most of the work has been done by its Indici support team and clinicians.

That’s partly due to a desire to ensure there’s no single person risk around the service, but also because ProCare found that it was easier for someone experienced in the health sector systems and processes to learn RPA technology than it was to teach systems and processes to someone experienced in the technology. 

“The main knowledge about what the process was was far more important than technical knowledge about what UI Path did.

“The PMS is a niche product and the clinical process is a little unique to the health system and even to the New Zealand health system or sub-regionally.”

Into the danger zone?

Roseman admits people are ‘understandably’ nervous about the risks of deploying bots.

“We know humans will be way less forgiving of mistakes made by a computer than they will be of the same mistake made by a human,” he says.

“So we are pretty cautious in the interests of a long time acceptability of the robots to take a measured approach to how ambitious we get with what they do.”

Among the terms and conditions practices deploying the bots are required to agree to are an obligation to review the bot’s performance, and to have a backup procedure in the event the robot is unavailable, ensuring there’s no excuse for things not to be done just because the robot is offline. 

“There are a lot of checks and balances as well as all the checks and balances that go into the clinical validation of what the robot does from inception and designing it’s decisions to testing it before release and then testing it at each implementation into each practice,” he says. 

Despite perceived risks and challenges – ‘obviously there is a degree of investment that goes in’ – Roseman is clear: “We are of the view at this stage that the benefits easily justify what you might think of as the challenges of putting RPA in place.”