Health and Sport Committee

Meeting date: Tuesday, February 21, 2017

Official Report 627KB pdf

---

Scottish Primary Care Information Resource

Good morning, everyone, and welcome to the Health and Sport Committee’s fifth meeting in 2017. I ask everyone in the room to ensure that their mobile phones are on silent. It is acceptable to use them for social media, but please do not photograph or film proceedings.

Agenda item 1 is an evidence session on the Scottish primary care information resource—SPIRE. I welcome Richard Foggo, deputy director of primary care at the Scottish Government; Libby Morris, lead general practitioner on SPIRE; and Scott Heald, who is from NHS National Services Scotland and SPIRE. I invite Richard Foggo to make an opening statement.

Good morning. I am grateful for the opportunity to update the committee on SPIRE, which is a significant part of our plans to transform primary care and general practice. SPIRE will provide a service to GPs and to the wider healthcare community for reporting on and extracting data from GP records in Scotland in a way that is safer and simpler than current arrangements. It has been designed and developed in full collaboration with GPs—through the Scottish general practitioners committee of the British Medical Association and the Royal College of General Practitioners—and in collaboration with patient groups and the Information Commissioner’s Office.

This committee has expressed a strong interest in how data and intelligence can be used safely and more effectively to plan, co-ordinate and deliver services, which will ultimately improve outcomes for individuals, communities and the whole population. GP-held data is potentially the most comprehensive data source that we have on the Scottish population’s health, but current arrangements do not allow us to maximise the opportunity to use that data to benefit patients and the wider population.

Current arrangements do not meet the most up-to-date data protection and information governance standards, and they involve different systems of variable quality that are technologically outdated and which can add to the burden on busy GPs. SPIRE is designed to address those concerns and to provide a safe, secure and simple system that can be more easily integrated with and linked to other data systems, including SOURCE, which covers wider health and social care data.

The most important questions that people have about SPIRE may well relate to concerns about privacy and the safeguards on confidential patient data. Robust information governance is critical to SPIRE, which has been developed in strict compliance with existing legal frameworks and best practice, as set out in the recently revised recommendations of Dame Fiona Caldicott. We have provided to the committee a briefing note on the information governance framework for SPIRE. The full details of the framework are set out in an extensive privacy impact assessment, which we will make available to the committee after this session and which has only just been published.

One of the most important information governance safeguards that we have put in place is to ensure that every patient has the right to opt out of patient level or identifiable data being extracted using SPIRE. Even when such data is extracted, significant safeguards will remain in place. However, to maximise public confidence in SPIRE, it is vital that every patient can exercise control over their data and how it is used. To ensure that every patient in Scotland is aware of SPIRE and of their right to opt out of it, we will launch an extensive public information and engagement campaign on 7 March.

The campaign will cover the radio, newspapers and social media but, critically, it will also involve GP practices engaging in an informed manner with their patients. No data extracts will take place until the campaign and that engagement have been completed. An important purpose of being at this meeting is to ensure that Parliament is informed of that development and that members have the opportunity to understand what is planned, so that you can all engage with your constituents with as much information to hand as you need. Following the meeting, we will send information to every MSP, and we are happy to brief you individually or collectively on any of the details that we discuss today.

Thank you. When you briefed us informally before, a number of us had concerns—particularly about data protection and sensitivity—and that is why we wanted to have a public session that puts everything on the record. That is the reason for bringing folks back.

Richard Foggo explained fairly well why it was decided that another approach was required. I would like to learn a bit more about the decision in 2013 to stop collecting practice team information. Has ISD Scotland or any other body continued to collect comparable data, even if it has not been published since 2013?

I am from ISD and I was involved in the decisions about PTI. PTI was a good start, but SPIRE is a much better and much more robust replacement. A key aspect of PTI was that it was limited to 60 participating practices. The idea was that those practices were representative of Scotland, so we could get an idea of what was happening in primary care across Scotland. However, PTI did not allow awareness of granularity below the Scotland level—at local level or within practices. We therefore decided to stop PTI and to focus on developing SPIRE. The utility that SPIRE will allow will be a greater ability to extract different types of data, to run analysis locally within a practice, which PTI could not allow, and to work at cluster level with the data.

SPIRE will be a direct replacement for PTI. The data that we used to collect from PTI could be extracted from SPIRE, if it was decided that we should do that.

None of that means that we have not published any primary care data in the intervening period. We recently developed a series of primary care indicators, which are published on the ISD website and which allow analysis down to individual practice level, which was never available through PTI. That series pulls together a variety of sources, including data from the old quality and outcomes framework. The intention is that equivalent data will be able to be extracted through SPIRE.

It has taken four years since the previous system ceased to get the replacement up and running. Have we still collected data during that period?

We have not collected data through PTI, because we decided to stop that, but we have continued to collect the QOF data. There are other sources of data relating to primary care that we have used in the primary care indicators—for example, prescribing data is available and has been used. There is also data relating to secondary care, which includes information about referrals from practices, so that means that we can publish equivalent data. That is all available publicly.

The replacement system for PTI was initially expected to be operational in 2014. Why has it taken so long for it to be operational?

The key point is that SPIRE is a more complex arrangement. One of the key issues has been agreeing the information governance principles, which Richard Foggo said were key to SPIRE’s development. Before we developed the software that will work in practices to extract and report on the data, it was key that we developed the information governance principles with key stakeholder groups such as patients, the BMA and the RCGP, which we did. We achieved agreement on those principles, which then allowed us to develop the software to be used to extract the data from practices.

Back when SPIRE was first thought of, the intent was to go for a big-bang approach whereby we could deploy the software to all practices and switch it on in one big bang. However, as the project developed, we recognised that a more effective approach would be to introduce SPIRE board by board across the country. Each board has a different set-up for how its GP information technology is provided and run, and it is critical that the SPIRE software does not interfere with that in any way.

A two-stage approach is taken whereby the SPIRE software is deployed to the national health service boards, which deploy it to their practices. We then make sure that there are no technical issues with the software, which is what has been happening over the past year or so. There is then a switch-on period, when we switch on the software, which is supported by training for practices to use it.

The project developed over time to ensure that it was as robust as possible and that we could deal with challenges as we went along. We also had an e-health assurance review—I think that it was last year—to look at what we have done and how we did it. It said that changing from a big bang to a more incremental approach was the right thing to do, and it supported that approach.

Is using SPIRE compulsory for GPs? I ask you to expand on that.

SPIRE is not compulsory—practices can choose whether to switch it on. Scott Heald explained how it will be installed in practices, but whether to switch it on is up to them.

There are lots of advantages for practices, as they can use the SPIRE software for their own purposes. They can use it to look at their information, run reports and examine interesting clinical questions, such as multimorbidity in their practice.

There is a second-stage process whereby patients can opt out of their identifiable information being extracted, if they want to. Anything that is anonymous is okay, in the old way that the QOF extracts could be taken.

Given that we will have an opt-out for patients and it will not be compulsory for GPs to participate, health boards across Scotland will have local variations and potentially a postcode lottery in collection. How will that affect the quality of the information for the Government and other organisations?

We expect that the vast majority of practices will want to sign up, because SPIRE will be useful for planning their own services. Scott Heald mentioned cluster working, whereby groups of practices work together—that is one of the new ways in which the NHS in Scotland is going forward. As a cluster, practices might want to look at a topic such as prescribing safety, in order to look at the services and consider such things as whether to work with a local pharmacist. We expect that there will be enough advantages for practices that they will want to switch on and join in.

We have not yet settled the new GP contract. Elements of the SPIRE extract might well have to be compulsory under the contract. However, that is not yet settled.

I add for clarity that SPIRE does not represent data collection; it pools all the data that a GP practice already collects. Even if practices do not participate initially in SPIRE, the underlying data that SPIRE uses is still recorded in the practice as part of its day-to-day running, so the issue is not about GPs choosing not to record the data.

I see huge opportunity from the project to improve practices—nothing is more powerful than data for healthcare improvement—and to improve the quality of research that goes on in Scotland. What concerns me is public perception, because there is likely to be concern about confidentiality and privacy. I would like to hear a bit more about your plans for public engagement to allay those fears and concerns and to help the public to understand what a powerful tool for improvement SPIRE might be.

The public information campaign is all about being open and transparent and about giving people as much information as we can about how the system will work. What is probably more important is all the patient stories that are behind the data. The Farr institute of health informatics research has a section on its website with 100 patient stories of how health has been improved by the use of research, and it gives practical examples.

We hope that other topics will become patient stories. For example, the news last night talked about astonishing differences in outcomes for cancer for people who live in different postcodes in Scotland, but the news just gave stark figures. It would be nice if we could use some of the SPIRE data to delve deep down and find some of the reasons for the discrepancies, or if we could use SPIRE to look at such things as how many patients are smoking, how many are obese, what drugs they are on, what other morbidities they have or whether they have mental health problems. Maybe SPIRE will give us some clues as to where we might find solutions to our problems, rather than just getting an awful headline in the news. We are building up a series of patient stories.

How long will the public engagement campaign that you are launching on 7 March run for?

09:45  

The campaign is due to run for four weeks. As Richard Foggo said, a variety of approaches will be used, including radio and newspapers. We have an extensive public-facing website that explains everything in plain English as much as possible, and there is a more technical website behind that for people who want more details. The privacy impact assessment that Richard Foggo talked about goes into great detail about how it is all going to work, so that those who want to read and understand that can do so.

We will have a helpline in place for the duration of the media campaign, which people will be able to call if they have queries. We also have a dedicated SPIRE mailbox that we will monitor throughout so that we can answer questions. The SPIRE team will be on stand-by throughout the period to deal with queries as they come in.

It is worth saying that we are mindful of public confidence, but we see the development of public confidence as a process and not an event. This is a conversation in which Parliament must have an appropriate role.

We expect the public information campaign to reach about 93 per cent of the Scottish population. We have put in place measures to ensure that we get as close to 100 per cent as we possibly can. However, very few campaigns are able to reach 100 per cent of the population. To go back to Scott Heald’s answer to a question from Alison Johnstone, we have spent a considerable time looking at the hard-to-reach groups and making sure that we will cover those who might not be able to use social media or who do not listen to the radio.

Underlying all that is the role of general practice and general practice staff in engaging appropriately with their patients. All that the public information campaign will do is make people aware that the conversation needs to happen but, ultimately, patients and general practice staff will need to engage in that conversation and people will need to understand that they have the right to control their own data.

I am looking to understand the current situation. If a practice knows how many diabetics it has on its books, for example, can it release that information without the consent of the patients who suffer from diabetes?

There are a lot of ad hoc arrangements across Scotland. For example, health boards might want to look at the information that you mentioned, and there are various ways of anonymising that data. Strictly speaking, the law and the GP’s duty of confidentiality say that information can be used for anything other than direct care only by taking out all the identifiers, such as date of birth, name, address and postcode. We can use machinery to scramble the information and in effect anonymise it, which is what SPIRE will do.

The problem for GP practices at the moment is that lots of people are asking for their information and other data. That might be for good reasons, such as cancer research, but they still have to stick to the rules, which are that identifiable information should never leave the practice without the patient’s consent unless it is for direct care.

I am trying to understand whether that will change under SPIRE.

There will be a uniform mechanism so that the public can be assured that their information will be stripped of all its identifiers. The process will be applied uniformly across Scotland.

Under SPIRE, if I was a diabetic, would I have to give my GP surgery permission for the fact that I am a diabetic to be used?

Yes. We are talking about the opt-out in relation to patient-identifiable data. Aggregated data in which people cannot be identified is not subject to the same approach.

Scott Heald might be able to say more about how the data is collated at the practice level, at a cluster level and at a locality level, but the aggregated data that does not identify people—they are just part of a group—is dealt with differently.

We certainly expect the majority of extracts that we run through SPIRE to be at the aggregate level. That is more about numbers and getting a sense of what is happening across the country.

If data is brought into ISD and we are feeding back data about diabetics for the country, we use other techniques. There is a process that we call disclosure control. We look closely at the tables that are published to ensure that there is no way of identifying individuals from them. Data is heavily scrutinised at all levels of publication.

I just observe that, if we are talking about public confidence in the system, such things need to be made clear. Permission is needed only for identifiable data.

Yes.

What is your definition of identifiable data?

There are two levels. One is individual patient-level data that has been aggregated so that it is at a high level and people cannot be identified in it. There is also data that might be at an individual level but from which we have stripped out all the patient identifiers. A system called Read coding is used in GP practices to identify the disease type that a patient has. From that, we can extract patient-level data with all the identifiers stripped off. Through SPIRE, with appropriate approvals, we can extract confidential data, which includes things such as names and addresses, but the information governance principles require the patient’s approval for that.

Is that additional approval?

Yes.

I understand that patients will have to opt out of the system. Everyone will be covered by it but, if I do not want to be on it, I will have to say to my practice that I do not want to be included.

Yes.

Good morning. I have a couple of questions. First, on the voluntary nature of the project and the uptake by GP practices, I am interested to know what sort of market research or user relations you have had with GPs. What do the profiles look like for those who are really excited about the project, those who are ambivalent and those who are quite reluctant to join it, and what are the reasons for the different responses?

We have worked closely with the Scottish general practitioners committee, BMA Scotland and the RCGP; they have been reassured by all the information governance, and they are supporting the project and recommending it to their practices. The RCGP in particular is concerned with the need to maintain high-quality services for patients and sees the project as a way to promote quality. BMA Scotland is more concerned with security and information governance, and it sees the new process as better in that it is open and transparent to patients while providing reassurance to practices that they do not have to worry about the extracts.

To repeat a previous point, the cluster working is a contractual obligation. Although we want to maintain a sense that the approach is voluntary in order to instil confidence, we also want to make sure that we have the data that we need to ensure that services are provided appropriately. That issue will be resolved in the next year.

My second question is on the application of SPIRE beyond health boards and the health sector. My colleagues and I consistently highlight the impact of planning decisions on local healthcare services. How readily available will the data be for local authority planning committees to interrogate when they make significant decisions on things such as the impact of large-scale housing developments and the siting of care homes, particularly in areas with an older demographic? How easy will that be, and how much foresight has been given to the application of the data in that way?

Essentially, the system that we have developed is robust enough to cover as many asks as can possibly be thought of. Richard Foggo mentioned cluster working—practices work closely with integrated boards on how they are planning and delivering services, and the tool called SOURCE to which he referred pulls together data to support the integration agenda. One of the big gaps at present is primary care data, and SPIRE gives us the opportunity to address that. The beauty of the system is that it is flexible, as it is designed to be used in bespoke ways for particular purposes. As those purposes emerge, we can develop the extract that will pull out the information as approved through the SPIRE system.

We are short of time, so we need to crack on.

I thank the panellists for coming along.

I want to touch on a couple of issues. I am still not 100 per cent clear about the identifiable tag. As I understand it, everybody’s data will go in and be aggregated, and people will not opt out of that, so what does the opt-out apply to? I assume that people’s personal data would not be identified anyway. What would people be opting out of?

I will give an example. In an individual patient’s record, there are fields such as postcode, which we would often use to derive the deprivation index. The deprivation index is not in the GP records, so if we wanted to do analysis to address inequalities, we could—with approval—pull out an extract at an individual level that would include some of the key diagnostic information and the postcode. We have a process whereby the identifiable data and the payload—the stuff that is used for diagnosis—are split; they are never brought together. That allows us to derive the deprivation scores, which we attach to the payload data, while we get rid of the identifiable stuff. In essence, that information is not identifiable, because the analysts in ISD would never have sight of any identifiable data. That allows us to use identifiable data to make those derivations.

As far as the opt-out is concerned, patients could opt out of having their data used in those extracts that come through to ISD.

I am struggling with that. The data in question is not identifiable as a result of the process that you put it through, and it is valuable from a health inequality point of view. Why would people be able to opt out of that process?

Under the principles that we have developed, fields such as postcode are deemed to be identifiable because, in effect, they identify where the patient lives. Therefore, we are giving patients the ability to opt out of that process if they wish to do so.

But they would not be identifiable from the analysis that you do.

That is correct, but it still uses the individual-level records. As Libby Morris mentioned, with regard to developing the information governance principles, some people still have concerns about extracting that level of data.

But, given that you have said that it is not identifiable at individual level and that the data is extremely valuable in enabling us to understand various impacts, including social deprivation, is your approach too belt-and-braces? Are you being over the top?

We are starting from a position in which we are trying to build as much confidence as possible in the system, so we want to extract as much risk as possible. You are right that we are being cautious. There is absolutely no doubt that, when it comes to patient confidentiality and privacy, we take a very cautious approach. Over time, we will develop the system and, as confidence in the system develops, we will look again at the principles that underpin it. We strongly believe that we should start from the most cautious position to build confidence with the public. That general approach is supported in the debate by the Caldicott principles. At that point, we will be able to look again at how that data is used.

I am concerned that you might be raising spectres that do not exist, which could be counterproductive. You are creating an impression that there is a problem when, from what you have said, there clearly is not.

There is a very delicate balance to be struck. We do not want the public to be overly concerned about a change in how their data is used. Once confidence has been instilled in the system, we can look again at whether we have been a bit too cautious.

The main issue here is the benefits, on which we need to focus. We talked about the health benefits; it is clearly possible to slice and dice the data in order to understand the correlation between different factors and outcomes. It is clear that GP practices can figure out where they are doing well and where they are not doing well. I understand that aspect, which you have discussed.

I would like to understand how far you plan to go on third parties. I am referring to people who are doing university research in areas such as cancer. A range of people, including some private sector organisations, will want to understand some of that data. What is the mechanism for interacting with those organisations? Have you given any thought to that? How do you plan—particularly in the context of the private sector—to generate some revenue from that area? The information that you are sitting on could be extremely valuable commercially. Big data analytics is an innovation focus for the Scottish Government, given its investment in the big data innovation centre and so on. That is exactly the kind of area to look at.

We are in the close-to-unique position in Scotland of having five million datasets that we can pull into one place and run big data analytics on. Are you conscious of the potential to use analytics from an economic development point of view to support the big data innovation centre drive? What discussions have you had around that?

10:00  

Yes—we are having detailed discussions with the Scottish school of primary care, which encompasses all the university research departments, and with the Farr institute to develop exactly the kind of informatics institute that you are talking about. Initially, we will focus on services for the practices themselves, but over the next two or three years we want to develop exactly that kind of facility.

To be absolutely clear, with regard to the commercial opportunities, SPIRE data goes through the well-established research principles that govern all health research in Scotland. Direct access to the data is not available to commercial interests; it would have to be part of a well-established research project with a partnership that is understood within the appropriate research framework.

It would have to be contained within an ethical framework.

Absolutely. The data is not for sale.

GP practices are going to get SPIRE. What about GP out-of-hours services, which operate from 6 o’clock in the evening to 8 o’clock the next morning and for 24 hours at weekends? Will they get the system?

At present it will just be the GP practices, but there are all sorts of plans for how we will develop in the future. Once SPIRE is fully rolled out to the practices, such services will be one of the next considerations, in addition to community nurses and—as we mentioned—social care.

One of the questions that no one has asked is what the cost is. What are the set-up costs for the system, and what are the annual costs of running it?

I have those figures. The set-up costs, which are primarily about developing and building the software, are £676,000. The annual cost to run the system is of the order of £400,000.

To put that in context, the annual cost for the PTI, which was mentioned earlier, and which covered 60 practices, had an annual cost of about £376,000. The utility that we will get from SPIRE for a similar sum of money will be far greater.

So it is not millions, then?

It is not millions.

I have one more question. The GP can get the information, but will it also be held centrally somewhere?

Based on the principles that we are building, extracts will be taken for a particular purpose and held for a time-limited period for that purpose. The data that is extracted from the GP practice will be taken into a safe haven in the NSS where the analysts can analyse it. Once that work is done, the extracts will, after a certain period, be destroyed.

That is fine.

Is the project on budget?

Yes.

Even despite the delay?

Yes.

A miracle—that is unheard of.

Will all patients be written to about the system?

They will not receive individual letters. They will hear about it on local radio and read about it.

I do not have a radio.

There will be information at the local library or in the pharmacy.

I do not go to the library.

There will be information at the GP practice.

I very rarely go to the GP. How do I find out about it? I do not think that what you have said is particularly acceptable. If we are going to make such a change and we want to instil confidence, people need to know about it.

You may put up posters and put on radio programmes, but, from my experience of working in my community for the past 12 or 13 years, if you do not stick stuff in people’s faces and make them very aware of it, you will miss a big chunk of the population. I have real concerns that you are not going to alert people individually that the change is going to happen. That has the potential to undermine confidence in the system.

We have looked into that aspect. Previous whole-population mail drops have not achieved the extent of coverage that you might have expected them to, but—

Maybe not, but people got the information.

It is a difficult question. Mail drops are very expensive for the benefit of reaching that final 6 per cent of the population who would not be reached by all the other means.

Presumably it would be a spend-to-save investment. We are not implementing the system for fun; it is designed—we hope—to improve healthcare, and I would expect that, down the line, it would save money.

We have made a judgment. In the early days of the public information campaign, there were a lot of discussions. We consulted widely and took advice from patient groups, and the decision was taken to do it in the way that we are doing it. I agree that we could have done it the other way, but that was what was decided.

I thank the panel for an interesting session. I want to pick up on the convener’s point. You say that you will be advertising on radio, in newspapers and through flyers in GP surgeries. How will you reach those who do not speak English as a first language or those who do not read?

The leaflet has been translated into nine different languages, including a British Sign Language video. We have also engaged extensively with the health and social care alliance Scotland and with learning disabilities networks to ask them how they would like information to be given to their particular groups. There are different methods—for example, we have engaged with Deafblind Scotland, which is producing its own materials. We have tried to tailor the material to suit certain groups, including those that are hard to reach.

My next question will be brief because I know that we are short of time. You say that the information will be advertised in GP surgeries. Is there an onus on GPs and their staff to alert the patients who are registered at their practice to what is being rolled out and the date when that will happen? You said that there will not be a mass switch-on—how would I know that my practice was signed up to SPIRE on a particular date?

People can opt out at any time—they can opt out now, even in advance of the public information campaign. No extractions will take place until at least eight weeks after the public information campaign; there will be no switch on of extractions for quite a while.

GP practices might use their own information for their own purposes on an on-going basis. They do not need to inform their patients about that, because they would be looking at the information that they have on their own patients.

We have also sent each practice a toolkit that includes many materials for them to use in informing people about SPIRE. We have been proactive in engaging with each practice to make sure that they have the information that they need. As Libby Morris mentioned, the leaflet has been translated into nine languages. All the materials are translatable into other languages, so if we have not covered all the needs, we can turn that round quickly—we have the steps in place to do that.

When is the system due to go live?

The switch on and the extraction of data will take place around 7 May, eight weeks after the public campaign starts on 7 March.

How many patients have opted out? How many patients do you expect to opt out?

No patients have opted out yet—the public information campaign will tell them that they are able to do that. It is difficult to answer your question about how many people we expect to opt out until we run the campaign. We have an idea, based on what the opt-out levels have been like for previous systems, such as the emergency care summary system. Libby Morris was heavily involved in that campaign, so perhaps she can say something about the level of opt-out for that system.

That was 10 years ago. Patients were leafleted and, in the first few weeks, several hundred opted out. I think that the total was 2,500; that number has stayed static for years.

Thank you for your time this morning—your evidence has been very helpful.

I suspend the meeting briefly to allow for a changeover of witnesses.

10:09 Meeting suspended.  

10:12 On resuming—