I want to highlight this as an important topic from a primary care perspective.
I dream of a future where referrals are made & automatically tracked from within a PMS (or equivalent), where patients can view the status on their online portal, and PMSs can selectively flag (through API updates within the PMS) only relevant issues (e.g. delays) to the referring clinician. As far as Iām aware this is not yet happening. Bring it on!
Secondary care providers also refer to other secondary care providers. And also sometimes to GPs (i.e. patient found to be significantly hypertensive perioperatively), or people in other parts of the country (or even overseas). Further complicating this is the public / private divide - it isnāt smart from an overall system perspective to silo referrals between them!
Any effective system has to be designed to accommodate these multi-directional, cross-sector, and cross-border aspects, and be broadly accessible by providers and patients.
Iām familiar with ERMS from a Canterbury primary care perspective - are you saying this solution is too siloed at the other end? As opposed to a semi-successful modular solution that could be better integrated into a PMS?
Iām only peripherally familiar with ERMS Iām afraid - but suspect that it was designed with a fairly tight scope of Primary ā Secondary care (which is hard enough). It does cover private and public secondary care, which is a major bonus though!!
Iām not sure that the order has a lot of meaning regarding priorities. But yes, point noted!
Complex topic for sure. I have some visibility from an Orion eReferrals perspective and related functionality. Our solution is deployed in both Northern and Te Waipounamu regions. In the south it manages triage of inbound (from primary care) referrals as well as all internal referrals (across secondary locations).
We also were one of the parties involved in supporting referrals for the Female Pelvic Mesh programme requiring collaboration (facilitated by HNZ) with ERMS, BPAC and Healthlink to integrate referrals (incl patients self-referring to the service) into 2 hubs (one northern and one southern). In order to support this all the providers of referral solutions noted had to integrate with the PMS vendors (at least Medtech and Indici, maybe also My Practice and Intrahealth).
Lastly on related requirements for appointment booking and viewing HNZ ran an ROI to check whatās in the market. This closed early Oct, and was expected to progress (demoās of what was proposed, followed potentially by a closed RFP) with implementation commencing February 2026. To date there has been no update since the ROI closed so I am not sure where this now stands.
The ROI was run via GETS so the docs incl requirements etc should be available in the public domain or from HNZ (although since it closed they may not be prepared to release them (or require an OIA request). I have the ROI docs so if anyone is interested please get in touch via chat and we can talk.
Central and Te Mawa Taki have not made as much progress on referrals (into and intra secondary care) to my knowledge but I am hoping we see progress in the coming year as it is a key enabler to targets (like faster access to cancer treatment, shorter FSA waits etc).
Thanks for the chat last week Paul and Nathan to help me clarify some of my thinking about this topic. I gave my undergrad digital health students an assignment last semester to create an innovation in response to the governmentās targets. Overall, they did a great job, bearing in mind that only a small percentage of them had been to see a GP or been in hospital (as patient or visitor), and had no experience from a clinicianās perspective or in writing code.
Here are some of the assumptions they made (Disclaimer: This is my summary of their assumptions with no emphasis on one or another assumption. I have not shared detail of their creative ideas because they own their ideas. The comments that follow are my take on the studentsā innovations). Not all of the students chose to write about referrals.
AI can be used to reliably triage referrals in a trustworthy manner. What they missed (and one would expect inexperienced people to do so) was the nuances required in quality referrals and the complexity of clinical decision making.
Referral tracking (as pointed out by @mca above) makes referral progress visible and should be done for (1) referrers, (2) those who receive referrals as part of their job, and (3) the patients and their whanau. Patient inclusion in tracking adds complexity for referrers and specialist clinicians because of additional workload to respond to queries. If something like a dashboard could be used to give an overview of a referralās movement through its lifecycle, there might be opportunities for more efficient workload management and referral process activities.
National tracking of all referrals was suggested by some students, but not in the sense of league tables. Rather, national tracking could give clinicians access to referrals that track from centres that are not in their usual catchment area. This would require interoperability on a national scale, and standardisation of referral structure and content (would it rob referrals of nuanced content that could have made a difference?). National tracking implies that population-based analyses can be done on the nature, type, and lifecycle of referrals, which adds richness to referral processes and outcomes.
What the students didnāt talk about in their assignments was
Using AI for a first pass quality check of referrals at submission point
Transparent connecting of different healthcare sectors, e.g., primary care to other services (not limited to hospitals) and vice versa
Using AI to allocate referrals to appropriate services (built into the software according to existing guidelines), based on content (this could change the routing of referrals, with associated risks of errors)
Ethical risks of AI getting triage decision-making wrong
The use of decision support systems to route referrals to appropriate services, do quality checking, or support a simple first-pass triage.
Two things that need resolution for referrals to work well (where some of it is automated) include (1) understanding of the typical referral lifecycle and its variations, and (2) interoperability (yes, that old chestnut) and associated standardisation in language, terminology, and format.
Great to see some fresh eyes cast on this issue @KarenDay Your summary points the way forward in separating the workflow and information concerns. IMO, we cannot make any progress without doing that and it might make it easier to gain the necessary consensus on a national solution. However, that still leaves us with the overarching problem in NZ Digital Health - the lack of a governance model for achieving lasting, non-political, consensus.
Really enjoyed your reflections @KarenDay and @Clax and also the follow up talking healthtech podcast captured some more detail too:
I think Iām interested in this from a primary care perspective because:
this is a big component of the admin work that GPs are often doing unpaid
this is an advocacy & equity issue - the work of tracking referrals is hard enough for high SES etc. patients - vulnerable patients are dependent on GPs to do this work so they donāt fall through the cracks
GPs are held legally liable if anything goes wrong in this process as the ācatch-allā person
When you look at it like that itās a pretty big issue, I think
Thereās a great deal that could be achieved I think picking up key elements from you @KarenDay and subsequent comments @pkjordan and @mca . I hadnāt listened to that podcast thanks!
Governance will be key to driving this forward and the optimist in me thinks we stand a chance (might need a tailwind). Streamliners and their Pathways adoption (ready baked governance) for out of hospital referrals is being enhanced I believe to understand AI implications/opportunities.
Iād argue the biggest challenge is the combination of AI, interoperability, and constructing trustworthy or reliable data for visualisation to both clinicians and patients. Both groups of stakeholders would love to see dashboards with work lists (for clinicians) and referral status incl predicted future timing/steps for (patient and clinician).
In terms of shift-left strategies in the context of workforce constraints and rising patient needs that in turn will need new āmodels of careā or ways to support more independent if not AI enhanced engagement that is trustworthy etc. This is massive and to date seems to have lacked focus from HNZ as the party responsible for defining the vision and strategy, but most importantly identifying tactic/actions to move us forward. Telehealth / virtual consults are one component that needs a lot of consideration of the broader context.
I agree with you @mca. Thing is, weāve had e-referrals in general practice since 2009 - have we mastered the basics yet? I hope so. The next step would be to link up all the referrers and create tools that make sense to everyone. Innovation can still happen in that space, but we need some governance (as @Clax points out) and then we can add layers of sophistication onto them. Whatās the next step?
The next step might be to engage the PMS and other referral application suppliers as nothing will change without their participation. I recall the first Referral, Status & Discharge (RSD) specification landing on my desk as a PMS developer back in 1997 - but we have definitely not mastered some of the basics yet! This is illustrated by reports from GPs (e.g. @richard.medlicott in the NZ Doctor this year) that RSD status messages are the single biggest cause of inbox bloat and - despite some worthy efforts (e.g. 2011 Pilot) - NZ is no closer to a standard for atomic discharge summaries than 28 years ago.
I love that. Youāre not allowed to change it to automatic (which I think is what you meant) haha.
Perhaps we need some research on this topic to create a tipping point. Or HiNZ could have a referral summit to gather everyone in the same room, or Health NZ could do that. Youāre right, I think that getting everyone into the same room with a purpose would be a good first step.
I meant electronic discharge summaries with structured (i.e. atomic) data.. The last NZ eReferral Summit Iām aware of was held at the HiNZ Conference 10 years ago. Suffice to say it was somewhat fractious, but things have probably calmed down since then.
In this context, atomic transactions and atomic data are two different things. Atomic data is that which is structured and rendered in a way thatās usable by machines for processes such as computation, comparison, optimal retrieval and statistical analysis - the antithesis of this is free text thatās usable by humans. The distinction is clear to anyone crafting or viewing health information exchange payloads, such as HL7 CDA documents or HL7 FHIR Resources both of which have distinct and separate areas for text/narrative and structured data.
ACCās Secondary Care management event yesterday identified problems with poor quality referrals as a significant contributor to the huge surge in the numbers referred to specialists for sprains and strains. I assume the new high-level model theyāre developing will reflect better referral processes.
Fair enough! Though Community HealthPathways is pretty helpful 
. The last NZ eReferral Summit Iām aware of was held at the HiNZ Conference 10 years ago. Suffice to say it was somewhat fractious, but things have probably calmed down since then.