PAS and pathology – the problem with old systems
- 9 January 2018
Many hospitals will have two systems at the core of their IT through nothing other than pure legacy. The first is a Patient Administration System (PAS) that largely came about to collect the NHS data model that arose out of the work of Edith Körner in the 1980s.
They will also have a pathology system that arose out of the fact that labs had analysers that were connectable and a naturally technically able workforce who wanted to join it up.
Both of these systems have in fact outgrown themselves for similar reasons. When people are doing things they look in the toolbox and see what they have. There’s a well known saying on this: if your tool is a hammer then every problem is a nail.
Thus, PAS has adapted to coding of the data for clinical and finance purposes, to run reporting solutions for analysis, and sometimes to take in departmental systems such as the Emergency Department (ED).
Pathology clearly had a need to do more than just report the level of serum rhubarb in a bottle of blood and their systems adapted to include a patient index so they could do delta checking and clinical reporting, that would otherwise be done in clinical hospital systems if we were designing the process today.
Along with these design flaws, the problem is exacerbated by another factor: we never renew anything. Therefore, the default position becomes to keep bolting things on in a most inappropriate way. This has the net effect of illogical system models completely killing the market, and we wonder why when you come out to buy something not much has been developed in the last ten years. Well, if you took out a system contract for 5+2 years and you’re still using it 20 years later, there is your answer.
Time for new systems?
So what would these things be if they were designed today?
A PAS is fundamentally a scheduler. I think we would go for a good scheduler from the market (there must be a market for such things). Nothing in this functionality would specifically need to be about health.
We would schedule everything from staff, buildings and equipment through clinics and patients. We would put the data into a warehouse for reporting and we would add the coding to that to have a rich platform for business intelligence, adding in other data for clinical decision support.
This would all mean we would not be encumbered by very niche and specific program add-ons for the NHS, making these systems difficult to replace. We would also not make this system look after the patient index part, another thing it has tended to become lumbered with.
Pathology would be a system whereby it would be able to manage a lab full of equipment dealing with a high throughput but essentially would not need to know anything about the patients. To look at range checking and other patient factors, the clinicians should be looking in hospital clinical systems (EPRs) for that, even the clinicians that are working in the lab, such as haematologists. If this were the model, the lab system would not even need to know a patient number. Much easier then to implement the seemingly favoured [Carter Report] model of factory style centralised labs. These things are now so steeped in tradition that they are probably difficult to shift until we can migrate to open platforms.
I should also add that of course there are large EPR systems that have been built from the ground up, but these are also getting long in the tooth and whatever your view on those, you should always have some kind of exit strategy.
Deconstruction
I have been considering how to go about this and along with our Global Digital Exemplar programme manager who came up with the term, and we are thinking about deconstruction as a way to get from A to B. It is going to be interesting to see if anyone in the market is up for this. Scheduling was after all a part of the clinical 5. Does anyone think they actually met that criteria?
The market
So how do we sort the market, this problem where we hang on to old systems and old models for too long?
I think the answer lies in the trend to turn all software into a revenue service. It has been too easy for people to make perpetual savings once a capital product has fully depreciated. It still works, so why invest in replacing it.
We need to keep the level of investment constant, and ensure that the platform is open so that we can change from one service to another at the end of the contract, or run multiple pieces of the jigsaw as separate services on the platform.
This will require a new approach from finance from top to bottom. We can see a looming problem with Microsoft around 2020 when our licences for MS Office terminate. . The same problem must also be addressed for all clinical and administrative software. The market must also balance to allow for the fact that the revenue is more assured; in other words, suppliers should not just receive a windfall on this.
Centrally, there continues to be a tendency to provide Public Dividend Capital (PDC) for initiatives where apart from the initial project costs, the overall “asset” is an ongoing revenue spend and service. This is also getting more difficult.
So there you go, I have thought about the directions to Tipperary and decided that I wouldn’t start from here. However, we do have to start from here, and we need a bit of imagination on how to get there.
23 Comments
Hi Ade, speaking as a Commercial Director on the supplier side, I whole heartedly agree with you that “suppliers should not just receive a windfall on this”. We, the NHS and supplier community, need to share the risk. I don’t want the NHS to “hang on to old systems and old models for too long” and that’s why I advocate that suppliers need to be able to offer the NHS a range of financial options, that will allow them to fund projects in a way that best suits their budget and financial circumstances. I have tried to address some of this in my latest blog http://www.imsmaxims.com/blog-and-opinion/no-cash-no-staff-how-can-the-nhs-afford-to-digitise/ Please take a look, and if you want to talk further, I’d be more than happy to discuss.
Deconstruction
Ade, I have long argued that PAS is over rated and probably not needed. Wirral first generation EPR (PCIS), did in fact not have a “PAS” as is commonly known, but rather built internally using the TDS toolset. This integration to the EPR brought data items together and pathway in some logical order, so clinical data was available for every order made to pathology. More on Pathology later.
We eventually built a Clinical Data Warehouse, and the DBA’s their did a excellent job in providing any number of slices and dices for reporting PAS activity, amongst many clinical sets. In fact Wirral in early days of RTT and clock stopping etc won a national award from Blair government as best of PAS in NHS at that time.
We had envisaged nurses being able to do admissions, discharges and manage waiting list. This did not work out so well, and data quality suffered. More coding tied down the loop holes and we passed on episode creation to Med Recs staff only.
The tight integration of identity management and episode scheduling was part of the clinical process and whatever the clinicians ordered/requested, lab were orders and x-ray/therapy were “requests”, guess you had to be polite to the non-lab people., and included relevant clinical context, FBC numbers were constrained by asking particular questions of reasons for test. Current results and outstanding requests also helped control numbers. Making it easy to order doubled workload figures until we tied down the pathways.
Later when I was a director of the Wirral informatics service, I installed the CSC IPM offering in community, and also learnt Lorenzo….
Well history probably says it all, but a lot of effort (for PAS) for very little clinical productivity, which is what the NHS is all about.
I did come across and formed a relationship (and friendship) with a small company that does scheduling, its light weight, agile, integrates and quick to install. (After hurricane Katrina in USA, they had it up and running for a large population in 48hrs.)
I have since installed it in Acute and tertiary settings for conventional PAS work and also advanced scheduling for Chemotherapy.
Give me a buzz and I would love to demonstrate how integrated, advanced scheduling engine, with integrated workflow and referrals, and ability for tight integration with telehealth, even to patient homes is available.
Pathology stirred my memories, I worked as a Chief Biomedical Scientist prior to my IT career (as many of us have), whilst the core of blood sciences nowadays revolved around robot tracks and large analysers, the number crunching for delta analysis needs computers to handle that workload. However, the QA process in labs is not just a analytical result, a clinical understanding of why that result is so high/low/abnormal is guided by the clinical context. Going back to EPR to look it up adds to time taken, forwarding the relevant clinical data with the order reduces this hit. Despite recent interventions, labs are not just result factories and have a role to play at the bedside. The role of integrated infection control further reinforces these boundaries, FNA in Breast clinics also demonstrates local pathology work near to the patient.
Best regards for you best of breed approach at Southampton.
Pete
Just a further comment on PAthology. It is a good idea , particularly for BoB sites to have a real enterprise MPI, I dont mean a PAS, but real population management across all your likely customers with all their different sources of referral. The other feature I would recommend is to “externalise” the lab/specimen number form the LIS, this allows you to build seamless pathways for workflows from GP , OPD, InPat, Community via a strong OCM system. Again Wirral WROCS, does that, you might want to talk to its author and designer, Steve Gerrard at Wirral, he is about to retire sometime, so sooner rather than later. You could even maybe tempt him to build you one. The GP’s love it, handles, Path.radiology,endoscopy, all referrals with bespoke pathways. The MPI and single source of the truth allows sharing across the community EHR both Hospital/GP data in the same record.
I also bid last year for some work for a University hospital of repute, one of the needs was to be able to order lab order and send (and get results) to any Lab organization in England. I came up with a design and possible product using some of the ideas above.
Ade, good points. I’m largely with you on that thinking.
May need a few hops to totally deconstruct and rebuild to get to core clinical systems functionality becoming ‘services’ used across the board.
We’re starting to look at wrapping next gen services around the legacy PAS that replace multiple ‘best of breed’ systems but still remains a best of breed for scheduling a best of breed for inpatients and best of breed for ED with shared services such as Order Comms. Removing OPD and ED altogether from the PAS.
Lab I hadn’t thought of how you put it – but I like it… certainly simplified things – there are some scenarios however where the lab will need the results for a single patient tied together and tie previous results together – Transfusion with blood types and transfusion history, micro and Histo where it’s far less automated.. So may not be able to get away from patient identifier entirely…. def worth the debate and kicking the re-design around a bit tho to see where those aspects lie and if it could be done differently
Pete. Thanks for those comments I will be in touch. Very interested in what you have said there.
Ade
The history of PAS systems tends to be somewhat unkind. The fact that 20 or thirty years later these finance focused systems are still in operation as a system of record does suggest the original design goal for the application was met. Replacing these non- patient centric solutions within a Trust is expensive and consumes thousands of days of effort to implement. Possibly 30,000 to 35,000 days between the vendor implementation support and the hospital staff effort.
The focus of solutions now needs to be much more patient centric. To support a new model of care, the patient needs to take greater ownership of their own condition and all of the appointments associated with that care pathway. The PAS needs to support solutions that are delivering online clinical appointment management for OPD appointments, for day ward management and for domiciliary care for patients with the greatest need.
9 NHS trusts have deployed this solution today as have seven of the largest Irish Acute hospitals. Not alone are DNA rates dramatically reduced but patients are seen at the appointed times and the stress in the workplace has reduced significantly. Waiting times have reduced and the patients can access appointments across multiple hospitals, planning their daily lives with the minimum of disruption. Almost 5,000,000 Clinical appointments were managed in the past 12 months using the Swiftqueue Enterprise Scheduling Platform.
Yes, this is one of those bolt on solutions that is delivering a patient focused solution. However, with a departmental implementation effort of an average of ten days effort it presents an opportunity to make a real difference in care delivery without expending a major corporate effort or cost. Today this is covering Oncology, Pathology, Physiotherapy, Maternity, Occupational health and shortly will include Cardiology and Radiology. All OPD departments, day wards and diagnostic departments can avail of this platform today.
Neil, Hmm, good points I think, especially Transfusion. Histopathology, yes and no. The reports will be in the order comms or whatever is handling reports. I’d expect the clinical staff to see history from there. It is however subjective to a degree, in other words, the previous report will maybe influence how you are reporting the current “thing”. I would not expect a clinician, pathologist, to be flipping between a lab system doing reporting, to a reports system, looking at history. It would not work. So I think that would be difficult to separate – so they need a link to the patient and patient history.
Is there a difference between things where I just tell you an absolute value, and things where I require judgement to report? Maybe that’s the answer?
Thank you for the comment on this piece. I expected it to be fairly controversial.
In fact I still hold with the “opinion”, but I also accept that not all order communications set-ups will be able to cope with this way of working, or at least would need modifying.
The point is that the lab really only needs to know the unique sample ID. It needs to accept this both on an interface and on the bar code for the sample, and match the two. It then needs to send results back to the originating system with that unique identifier. It does not need anything else.
Any clinical opinions should be given in order comms and other clinical systems, where of course the patient will be known.
I think this is a simpler model, though as I have said, it would require a huge change in both products and attitudes. I am up for the debate though, as this way of working certainly facilitates a more lean laboratory structure Carter style.
Ade
I think your post agrees with my view. I certainly wasn’t saying that PAS is easy to implement. I have done a couple and have the scars. What I do think though is that we would not, in retrospect, build some of the clinical functionality that has been built into some of those systems. I am interested in this Swiftqueue system. Where can I see it?
Ade
Hi Adrian,
Thanks for the response and for the post. I sold some of that stuff 30 years ago so I am guilty as charged.
We would be happy to show you our platform covering Online patient bookings, online registration, Patient flow management, Speciality clinic management, Day ward and domiciliary care. We might even get to show you how we are scheduling Telehealth resources in the US for a large Hospital group.
When and where might be good for you. We could look to visit one of the trusts in London or Coventry is that worked for you.
Lets have a call on the phone first and take it from there.
Kind regards
Noel.
I met Brendan Casey (CEO & Founder) at the beginning of this decade with regards on line Phlebotomy scheduling so patients could book their own timeslots. I was impressed with his foresight and planned to work with him in other areas of on line appointment scheduling but was derailed when our Trust was acquired. I have watched his progress on LinkedIn and his desire for simplicity in the solution has taken him from strength to strength. Unfortunately I haven’t had the opportunity to do any further business with him but I think they are a company to watch.
I agree with Adrian’s point about lack of innovation in the PAS and Pathology sector, and a tendency to just “bolt on” rather than sit back, take a wider view and ask the question “what do we really need this to do?”
However, as a former diagnostics IT manager, I have to disagree with this statement : “If this were the model, the lab system would not even need to know a patient number”. How would a lab system lookup patient details from an EPR without a lab number, and without a patient ID and a mapping in a lab system, how would a barcode on a blood tube get linked back to a patient ID so that results could be distributed and alerts created (where necessary) ?
I also dispute these statements : “We would also not make this system look after the patient index part, another thing it has tended to become lumbered with” and “Pathology … would otherwise be done in clinical hospital systems if we were designing the process today.”
These statements fail to account for the fact that over 50% of the work processed by most hospital labs actually comes from outside the hospital. The vast bulk of this is GPs but also edge cases including private requests from clinics, industrial occupational health departments and even private individuals. Is Adrian really suggesting that all these patients could be effectively registered on an acute Trust EPR and all allocated a hospital number.
Nonetheless, I do agree with his idea of interfacing to external sources for the demographic index. I would suggest that, given the NHS central investment in it, all lab systems should index to the Personal Demographics Service, use the NHS Number as their primary identifier and that acute EPRs translate NHS Number into Hospital ID when ingesting results for the cohort of Pathology patients whose requests originate from the acute Trust who host the laboratory.
Using the NHS Number would also deal with the bane of path system admin’s life administering duplicates entries / merging / changing IDs for pregnant women where some tests originate from the community using NHS Number, and some originate from the hospital, using hospital ID.
Sorry, if lab systems had to connect to the spine, the users, including the analysers, would all need smart cards as it stands. No way that’s going to happen.
Ade
Hi Ade,
Thanks for a very stimulating article. Just to put you right re: connections to Spine necessitating smartcards for users; it’s a bit more subtle than that. We have deployed PDS Spine Mini Services into a number of scenarios like this one, and indeed have even had discussions with Path System vendors re: automatically looking up records on PDS – all without the need for Smartcards, or compliance for that matter.
Paul
How do you cope with patients without NHS Numbers (Scottish, Irish, EU or unable to identify)?
An identifier is mandatory but a sensible choice should be allowed (PAS Number, passport, etc) for electronic interchanges. Also including English/Welsh NHS, Scottish CHI, Scottish NHS and Irish HSNI number(s) is highly recommended and must be done is they exist and have been verified. The official NHS identifier is indicated as such.
On paperwork just show the official English/Welsh NHS Number (if in Wales or England).
Thanks Paul, I know discussions have been going on, but as it stands I still cant do basic Spine look ups from my Intersystems TIE even though I actually have the SMSP interface, just because of somebosy’d idea of governance.
Ade
Tor, I’m not getting this. I need an identifier for a sample. In order to tell you how much rhubarb is in it, I don’t need to know who the patient is. It’s debatable?
Ade
Fairly trivial, I know, but sometimes there are cases like if the patient is under 18 measure custard as well as rhubarb – which may be set up in your order comms but is a lot easier if it’s in one place, your LIS.
Interesting article and reflects the way the LIS I work on is moving from being a system in it’s own right to just taking orders, sending them to a humongous analyser and then sending the results to the requestor.
Hi Ade,
I’m not sure what IG issue is preventing you connecting to the Spine via an approved PDS SMSP as we have deployed them for dozens of Trusts. Feel free to give me a call on 07801 106989; you never know, a 5 minute chat might lead to resolution of the issue.
Kind regards,
Paul
Ade, commenting on identifiers in general.
Re ordercoms, LIS, PAS, etc. Yes we need to deconstruct to put his back together.
I leaning towards LIS having open api’s which allow querying for past/current orders and reports. This would allow orders and reports to be shared to any HCp giving care – would assume this helps to prevents duplicate orders and get reports to all clinicians. Searching LIS should be easier than finding the EPR holding the patients reports?
My preference is for systems to allow (real time) access to the data they store. Then it doesn’t matter that much if your appointment, scheduling is in your PAS, observations in you pathology and EPR. You can still do best of breed, keep you PAS, as long systems provide open access. If your system doesn’t do this, should that be a sign it needs replacing?
Tor (and DCE)
Yes and yes really. I think the lab systems need to evolve to use open APIs so that they can retrieve data that they need. I knew this piece was not a cut and dried “we should do it this way” thing, so just progressing the thoughts a little further – the lab will know an identifier for the sample, and the order comms that sent the order will know stuff like how old the patient is. Should we a) send useful stuff like the age of the patient etc in the order message or b) should we have an open API so the lab system can query the original system and obtain it? If we sent enough information with the order, but still indexed it on order ID not patient ID within the lab, would that be enough and not require the complex patient database interactions between providers and potentially external lab services? Probably all getting too complicated for this forum now.
Wonder why the lab has attracted attention and not PAS?
Really good point Adrian.
My experience with the Pathology department and all of the disciplines suggests these clinical professionals tend to be “can do people” with very clear views on whats needed and what works. As a clinical department they tend to be very focused and in many ways quite black and white with very few grey areas going on. Once we land in the PAS. HIS, EHR, EMR world you land smack bang in the middle of a dulux colour card. Wonderful possibilities, a broad focus and loads of gotchas
I believe patients deserve better. Herding patients like cattle does not engender any level of commitment to manage “my” personal health issues. Not when I turn up at 8.30 for a 09.00 appointment along with fifty other people. When I am seen the file may not be available and I get invited to do “more tests” and come back when we may have found the chart. I think more can and should be done for Patients and leveraging from the system of record is where this effort should be placed. If patients were treated as individuals then getting a buy in to a new model of care would be a much more attainable goal. Domiciliary care is a good example of an area where digital health solutions would really benefit the patient and support the challenges facing the Acute hospital.
Regards
Noel.
Hello Paul [Richardson], thanks for that offer. I could get our SMSP person to call. However before I do. I think I may have misled you a bit. We can do a look up now, using the connection from Intersystems. What we cannot do via that is bring certain aspects of SCR across that we can populate in the EPR to stop clinicians having to log in [to SCR]
The lab system kind of works backwards as it maintains its own MPI and then checks out to see if there is a record to link to. We would change this if we replaced LIMS.
If you think you can help with the first one of those then we should have that call, and thanks for the offer. We have been discussing this with Intersystems for >1Yr
Ideally we would also like to be able to update Spine but I feel that is another story
Comments are closed.