19
Examination of witnesses
Dr Colin Campbell, Paul Midson and Simon Hailstone.
Q139 The Chair: Welcome back to today’s meeting. This is the 12th oral evidence session as part of the committee’s inquiry into childhood vaccination rates in England. Thank you very much to our witnesses, Dr Colin Campbell, Paul Midson and Simon Hailstone, for attending today. The session is open to the public and broadcast live. In a little while, it will be accessible also on the parliamentary website. A verbatim transcript of the session will be taken and it will be published on the parliamentary website. A few days after the session, you will be sent a copy of the transcript to check whether there are any inaccuracies or anything like that, and it would be very helpful to us if you could advise of any corrections as quickly as possible, so we can get it up on the website. If, after the evidence session, you would like to clarify, simplify or amplify any points that you make this afternoon, please submit some supplementary evidence to us. When you start your first answer, please briefly introduce yourselves.
We are talking about data this afternoon, so what are the key data systems for childhood vaccination programmes? What are the key strengths and weaknesses of those systems? In what ways do data systems for childhood vaccinations influence coverage? You might perhaps focus particularly on the forthcoming restructuring of NHS commissioning for vaccinations. We will go to you first, Colin.
Dr Colin Campbell: Good afternoon, I am clinical medical consultant at UKHSA, and I lead the team responsible for producing the national statistics on vaccination. On the key data systems, it is important to point out that there are numerous data systems and it is a complex landscape. We have data systems about where the data are held: whether they are in GP IT systems, in the child infant health information systems or in national datasets such as the data processing service that NHS England runs.
Then there are also the systems that we in UKHSA have to collect the data. We broadly collect aggregate data through the COVER programme, which has been running since 1987, and we also run a series of collections through the Inform platform, which are specified by us to pull data from general practice across a number of the collections. More recently, have record level collections, which came out of the Covid work originally commissioned for flu but then rapidly transferred into Covid during the pandemic. That has evolved, so we are now getting increasing amounts of data on a daily basis for all the Section 7A vaccine programmes.
Strengths and weaknesses was the next bit of the question.
The Chair: I was just going to ask: is the fact that there are numerous data systems a strength or a weakness?
Dr Colin Campbell: Both. For example, COVER has been running since 1987, so it provides a long-standing, useful timeline of data across the thing. It is the major dataset that produces our national statistics for vaccines up to age five, and it is also the data that feeds the WHO indicators that we provide. So, it is strong and long established, and the specifications are governed by an information standard—but it does have its weaknesses. For example, it is available only in aggregate form, which limits our ability to do things such as breakdowns for ethnicity—that is not available in COVER. We cannot do deprivation, except at a very crude local authority level, with that dataset. It does cover unregistered children as well: it is meant to cover that vulnerable group. It does not always do so in a consistent way, but we know that in the systems behind COVER, such as CHIS, which Paul and Simon know a lot about, there are certain difficulties and weaknesses in the data.
There is not always complete concordance between systems: the vaccine information that is held within general practice is not always exactly the same as that held within CHIS, for example, and that can cause problems. The Inform system, which we commission, has the advantage that we write the specifications, so we can define very clearly what we want to collect from the GP systems, and so we in UKHSA control the validation of that system. It also means that when things go wrong, we have communication to confirm and cross-validate that. But there are also lags in instituting that. Those specifications are reviewed annually, for the most part. We can write them to measure whenever there is a change in the programme, but it takes time for them to be implemented by the GP IT system suppliers for them to pull the data out, and that is a significant lag. But once the system is running, it continues to produce data.
I would say that the third group of record-level data that we have had since Covid does produce daily data, but of course it is dependent on the quality of the data that is held in the system from which it is pulling. A previous witness spoke about RAVS: RAVS will feed into that as well. As an example, likewise, the Mavis app that is being implemented by NHS England for school programmes will run into that.
These data are primarily held prospectively. So although we might have record-level data, say, for the new RSV programme, it is only available from the start of the RSV programme. That means, for example, that although we have record-level data for MMR, we do not have very good historic levels for MMR. For the new programmes that are coming on stream this year, we do not have data going back. That means we can look at what is happening now, but it is very difficult, in that particular dataset, to look at what happened five or 10 years ago.
Q140 The Chair: I see. Thank you very much. Can I move on to Paul Midson, please?
Paul Midson: I am head of child health informatics for the NHS South, Central and West commissioning support unit. I have been working in Child Health Information Services—or CHIS, as it is more commonly known—for the past 17 years. In my current role, I am responsible for the child health systems, data flows, reporting and analysis.
As Colin said, there are many different systems in play in childhood vaccinations. Obviously, I can talk mainly about the child health systems, which is essentially a destination where that vaccination record ends up, and it is a child health responsibility to collect all that information and share it with the agencies that need it.
The strengths of the child health systems are their configurability. They are highly configurable, allowing you to cater your call and recall for certain vaccinations—but that can also be a downfall. The more child health teams configure their systems differently, the more we end up with different results. One of the main weaknesses for us is not necessarily in the child health systems themselves, but in the data flows that go into them. Colin spoke about the data quality issues, and some of that can be driven by the lack of substantial data flows going into the child health system from GP practices or school immunisation services. As I said, that is not necessarily a weakness of the child health systems but of the wider system.
As for how the child health systems influence coverage, I think a lot of it is based on reported coverage. Again, it comes down to data quality: if you do not have good-quality data, you are not going to have good-quality reports and statistics based on that. Part of that can be solved with the flowing of the vaccination data. So integrated, automated data flows from GP practices into child health systems would reduce a lot of missing vaccination data. It is also about tackling the data quality issues when they are being recorded.
We have seen lots of cases where children will move into the country and the GP practice will record their entire vaccination history as being given on the same date, so you can see that they have had all the listed vaccines, but as a child health service, we do not know when they were given. We do not know if they were valid and what they should be given next. As I said, it is not necessarily a weakness of the system but a weakness in how we can report those figures.
Q141 The Chair: Thank you. Simon Hailstone?
Simon Hailstone: I am a public health consultant working for NHS England’s London regions, specifically in the public health and primary care commissioning team. I focus on antenatal newborn screening and on child health information systems as part of that role.
I agree with everything that Colin and Paul have said, but will mention some of the other, maybe lesser-known systems that support the wider running of the vaccination programmes, including the NHS Personal Demographics Service, PDS. That system allocates NHS numbers as one of its key functions and is therefore critical to our establishing the eligible population cohort that we are trying to reach through our programme’s efforts. Another system, the National Events Management Service, NEMS, is a tool designed not specifically for vaccinations but to relay information between systems and services, such as CHIS and the PDS, about patient movement and activity that has taken place in other providers.
I mention these because they add to that picture of complexity. NEMS is a tool to try to keep the system in sync with itself. There is a concept in system design, particularly in database design, where you have a single version of the truth. In vaccination programmes, we do not really have that structure. The truth is disseminated out there across GP practices and across different CHIS services, and we have a variety of tools that try to facilitate access, to both add information to those records and to collate and present it to the people who need to interpret it.
I would call it an “onion skin” system, in that, after the point of a piece of vaccination activity being coded by a GP, for instance, there is then a level of logic—an algorithm, say—that extracts it from GP systems into CHIS. At the point of CHIS loading it into its systems, there is another level of data validation and interpretation of what that means before it is pulled in, and then, on top of that, other levels of interpretation and extraction that feed, say, into the COVER statistics.
It is challenging. We are doing some work on this right now in London to better structure how we do this and try to dig into where errors might creep into that process. If you have an issue with coverage in a particular practice, for instance, it is not immediately clear what degree of that is due to population factors, which we have been discussing, or access issues, versus potentially real activity that has been either excluded from the record or interpreted in a way that does not quite fit with the specifications of the systems at play. It feeds into that interoperability question and is a wider weakness of that evolved system.
The Chair: Baroness Hodgson and Baroness Cass would like to get in, but I would like to quickly probe a few things first. Will the unique child identifier number be helpful?
Simon Hailstone: I think it will. However, it will almost certainly stumble across some of the same questions that we have started to ask. Colin and I have been doing some work on this particular challenge. I will probably annoy my colleague Dawn Hollis by citing her. She is an expert commissioner and, when I first started my role, she said to me that one of the main issues that we have in our systems is that, as a society, we have taken the decision not to know where all our children are, and that permeates our programmes. We do not know whether someone is not picking up the phone because they do not want to or because they are not there. Having a record of children is very much the problem that our colleagues in CHIS are trying to solve, in a sense. The NHS tries to come up with a proxy measure of the population, with input from the PBS and CHIS colleagues, and it will always be imperfect. There are some challenges in that area.
The Chair: We have heard that there are certain groups, such as Romani, Traveller and Gypsy populations, that we do not know much about because nobody asks. Whose job is it to ask for that sort of information so that it can be gathered? Is it the GP’s? What if they are not registered?
Simon Hailstone: Ideally, it would be through health outreach and GP registration, with practices working with those communities. It is probably a question that local authority colleagues—
The Chair: Is there not a box to tick for that particular group?
Simon Hailstone: We do collect ethnicity information, which, I recognise, is not quite the same. Work is under way to improve things like engagement, but the data collection on that group is variable across the patch, certainly in London.
The Chair: So in some situations it is not quite as granular as might be helpful. Three colleagues would like to ask supplementary questions on this: Baroness Hodgson, Baroness Cass and Baroness Andrews.
Q142 Baroness Hodgson of Abinger: Thank you all very much for what we have heard. Data is quite a generalised term. Are you able to put a bit more flesh on what you mean by data and data collection? Is it just quantitative? How personalised is it by the time it comes to the centralised systems? Do you get any qualitative issues coming out of this? If, for instance, there was an issue with a particular kind of vaccine, would this system throw it up? You talked about analysis. What kind of analysis will you do on this data?
Dr Colin Campbell: To answer your first question about what the data is, fundamentally, we are collecting information about the vaccine. Depending on the exact system, it will be, for example, the batch number, expiry date, the date the vaccine is given and the location in which it is given. That is where we have the record-level data. We do not have that in the aggregate counts from some of the other systems.
It is fundamentally about the vaccine, but we also we need to know about the people—the eligible group who are having the vaccine, so that we can know the denominator for that. That is also collected in different systems, either in aggregate as counts or, increasingly, as record-level data. That forms the main denominator file, and the linkage between that is the NHS number. What is crucial in that is having a correct denominator file, ensuring that everyone who is in that is actually in the country and should be counted and not duplicated. We deduplicate on the NHS number, but we know that there are additional problems with that.
In terms of the analysis that the data feeds, my team does the coverage—what proportion of an eligible group has the vaccine, primarily. You asked about problems with a vaccine. For example, if there was an adverse event that was unexpected, that would be reported to the MHRA, which might then consult us to undertake specific studies. We would use that data to undertake specific studies to investigate a safety issue, should that arise.
The Chair: Would anybody else like to add to that, or should we move on to the next question?
Paul Midson: I can talk a bit about the analysis side of things. The child health service is an administrative rather than a clinical service. We have a service specification that says what we should produce and what we are contractually obligated to produce and submit to the UKHSA. Colin mentioned the COVER report, which looks at a very narrow group of children: those who have reached one, two and five years of age in the reporting period. We have developed additional reports to supplement that, which look at the whole nought-to-19 population, and we publish those for NHS England, which commissions the service. That takes into account all the different factors that COVER does not; we look at things like ethnicity and deprivation to give a wider picture. It is not very specific; it does not say what we need to do, but we have taken it upon ourselves to do that because it gives the best-quality information.
The Chair: Will you give it to the ICBs—
Paul Midson: Yes, we also share it with ICBs and local public health teams.
Q143 Baroness Cass: This is probably naive and stupid, but I am going to ask it anyway. I had originally assumed that there was an automatic system whereby things got from primary care to CHIS, but you said it would be better if it was automated, so I construe that it is not. What does a GP need to do to make the data go from the primary care record into CHIS? Then Simon mentioned that there can be problems where it is kind of filtered or lost. I would assume that the information was a series of very simple data points that are pretty reproducible and universal, so why would some selectively get lost? Does that make sense?
Paul Midson: In terms of the data flow to CHIS? In a lot of cases, the GP practices themselves have to run system reports, from their own systems that they design, looking at, for example, all the vaccinations given in the past week. Oftentimes, they will send that into CHIS via email. We have had to go out to third-party suppliers to look at extraction methods to extract information from practice systems and feed it into the system without the GPs having to do anything. That has raised some concerns from some practices and ICBs about information governance. We have done a lot of work behind it, to get it to where it is now, and the systems that we use are in use within practices for other purposes as well, so the governance is there. I think there is just a bit of hesitancy, maybe—a bit of reluctance from practices who have not done that kind of thing before.
Baroness Cass: You said that it might come from email. It might come in various different ways to you and you have to extract multiple different kinds of sources?
Paul Midson: Right, and GP practices have different systems as well, so we in CHIS might receive it in three different formats depending on what system the GP practices use. We then have to transform that data into a format that we can upload to the CHIS system, so there is a lot of manual work there.
Baroness Cass: You are looking very youthful for having worked in this for 17 years with these challenges.
The Chair: Is there any guidance as to how often they are supposed to run that system? Sometimes, delay in receiving information is as bad as not getting it.
Paul Midson: Exactly, especially since CHIS is responsible for call and recall. How can we call children for the vaccinations they need if we do not know what they have had? We normally ask, if it is a manual system and they are submitting via email, that it is run weekly. There are very few practices that we deal with now where that is still in place: it is predominantly the automated mechanism that I spoke about. Previously, it was weekly, but we do get some practices that might forget—the person who normally does it is off that week or whatever—so there are inherent delays.
Baroness Cass: Can I ask Simon about the bits that get pruned out or lost?
Simon Hailstone: Absolutely. I would suggest, as one example, differential approaches to data validation, which is the checking of the integrity of data that comes through. Colin mentioned a couple of these earlier, but one example would be if, for instance, a vaccine comes through with a date saying that it was given before the child’s date of birth, you have two options. You either take that as being sufficient evidence or you do not, and that is a bit of a value judgment in the system, which is applied inconsistently at different points. Certainly, our CHIS tries to identify those issues as exceptions and work with GP practices to try to correct those issues, but that does not always happen.
On top of that, there are different interpretations of what should be included in the different specifications. We have mentioned COVER reporting. There are a variety of exceptions built into the COVER definition. One that springs to mind is that if an MMR dose is given before 12 months, it should be excluded from COVER reporting. There, we get into the underlying philosophy of what we are trying to measure in these programmes. Do we care most about just figuring out if anyone has any level of protection, or are we trying to understand whether a vaccine has been delivered strictly according to the NHS-defined programme? Those two things do not always agree with one another.
Q144 Baroness Andrews: I have two questions, one of which is rather massive, and I apologise because I am very diffident about asking it. The first question is: what did you learn? You were in the room when we were talking to our previous witnesses: what did you learn from what they were saying about gaps that you might not have appreciated, or the difficulties that people have just in the ordinary administration of accessing data and using it? Was there anything that surprised you, or did you know most of that?
My second question springs almost immediately from Simon’s last answer: are you all agreed that we know what we are looking for when we are trying to create the picture of the national health in terms of child health? Are you all agreed with the elements? Have we got all the elements covered? If there was an exam question, would you all write the same answer to that? What is it we are trying to find out? What are we trying to diagnose? Or are we trying to measure something, which is different, is it not? I know it is a broad question, but I am afraid I cannot resist it.
Dr Colin Campbell: If I may, I will start with what I learned from the previous witnesses. There were two things. One was about the availability of the summary care record, for example, to pharmacists. That is very much dependent on the quality of the data that is available to read. Some of the problems we are talking about are a little bit upstream of that: it is all very well having a summary care record there, but the data has to be correct to read that in. That applies across all the systems.
I think it was Caroline who spoke a lot about community participation and co-creation, which is really important. It just so happens that we have been leading a European project looking at improving vaccines in underserved communities, and the entire focus has been co-creation, with some incredible examples from the field—from the Roma community, for example, in Slovakia. There were very unexpected findings—rejecting an outreach, for example, because it reminded them of Covid. I was really pleased to hear her say that.
Baroness Andrews: And you can use some of her work, of course.
Dr Colin Campbell: On your second point—would we agree on the exam question?—Simon mentioned one view or one version of the truth. In some senses, I do not think that that exists. We have different questions to answer, maybe with the same ultimate source of the data, but, for example, if the NHS wants to report on how many vaccines it has delivered and to where, that is a slightly different question to what the coverage is in that population. When you are looking at the coverage, you need to take account of who is left—who is still alive—but you are not doing that when you are counting. You need to have the data available, to be able to look at it in different ways, to answer the questions that you are trying to answer.
Baroness Andrews: Fair enough.
Simon Hailstone: If I can come in there, I agree with what Colin was saying. The issue around the fundamental ask of what we are trying to do is extremely important. There are these two extremes. It is, “Are we doing an NHS programme that is designed to measure what an NHS programme has done?”, versus that wider question of total population coverage. They are two completely valid questions. Right now, what we are trying to do is a halfway house on both those issues, and we are misconstruing what we are therefore interpreting from the data. If we want to do both those things, which I feel is the sensible route on this, we need to facilitate and give the remit to the services to do that.
That is when you encounter these issues around, say, data sharing: if we want to maintain an accurate population register—a total population register—what does that mean? Presumably, to be somewhat glib here, it does not include children who have come here on the Eurostar for a weekend to visit London from Paris. It does include people who were born here, who have never moved house and have lived here for ever. In between those two extremes, there is a whole wealth of different life experiences, going via children who live transnational lives. We need to understand what the appropriate line is in order to answer the questions about programme performance and population coverage.
We need to talk about terms such as “ordinarily resident”, which we do not tend to do in vaccination programmes. We talk about children and give age ranges, but we do not put a line around that. It needs to be done carefully to put a tighter definition on the cohort, because only then can we start asking questions. If we think about “ordinarily resident”, that might mean one thing when we have to speak to colleagues in the Home Office, for instance, about ins and outs, visa stays and so on, to try to inform that view of an accurate population denominator, but it also spreads more widely.
In the previous session, there were some questions about the payment for vaccines. In my view, we need to pay for data completeness as well. It is about not just the NHS-given vaccines, where the concept is fairly straightforward: we say, “If you’ve given someone a vaccine, we will pay you for that activity”. We are asking for more than that, particularly in places such as London, where we are saying, “This person has been on the French vaccination schedule”—I am picking on France a bit here. We need to pay for that information to be complete.
Right now, we have a system where it is not easy to interpret the lack of information on whether a vaccine was given. We interpret “no information” as meaning “not vaccinated”, when it could mean, “We never asked in the first place”. That is quite an important distinction when we start getting into these knotty questions about what precisely we are asking the system to do.
The Chair: It strikes me that, if you are interrogating a particular dataset, you really need to know about the strengths of that dataset—and probably its weaknesses as well.
Paul Midson: Yes.
The Chair: Perhaps we should move on now.
Q145 Baroness Wyld: This flows quite well because my question is about improving data systems, including the responsibility and accountability for improving those data systems. Can you set out for us who is responsible for that? Within that, can you talk about the key priorities of DHSC, UKHSA and NHS England? In your view, what are the strengths and weaknesses of the approach?
For the second part of my question, can you then talk to us about the wider NHS data reforms that are coming up in the system, including the single patient record? How do you perceive the risks and the benefits?
Simon Hailstone: Sure; I will do my best.
Baroness Wyld: I know that it is a huge question.
Simon Hailstone: Exactly. I go back to the previous point about the complexity of the system, including the commissioning arrangements and the accountability for those different systems.
For instance, in London, my team are the commissioners responsible for the Child Health Information Services. We do not commission their IT systems; they subcontract for that. So we are already in an arm’s-length relationship with the system suppliers. Everything is underpinned by standards, but we still face and come across issues around the interpretation and implementation of those standards. I think that there is still a question around how we can better simplify the set-up there.
You mentioned the single patient record. Again, I think that, in general, it is a very good and positive thing, but it will add another version and interpretation of the truth when we have not gotten into the weeds—into the detail—of how we can make sure that we are building a consistent picture of the truth. I certainly get into the world of wondering about how we can make algorithms more consistent. If you are going to have a bundle of vaccination activity data and information and another bundle with bits of coding around declines, how do you consistently take that for a given individual and make sure that you are getting a consistent outcome from it?
In terms of positivity in the system, a lot of work has been done on improving those standards. I am not an expert on this but FireAPI, which is the standard for shuttling health information around the system, underpins a lot of things. However, there is still a level of friction, which is very difficult to pin on any single body in all this. It is one of the areas on which Colin and I have focused recently in order to understand it better.
Baroness Wyld: Okay. Can we go to Dr Campbell, please?
Dr Colin Campbell: Yes. We touch on all the same themes, do we not? It is about data system complexity and interaction. Obviously, GPs are responsible for the accuracy of their clinical record. Likewise, the CHIS have their own areas of responsibility, such as how they manage call and recall. Then we have the digital strategy set by NHS England. A lot of work has been done there to try to bring all that data together. Different organisations have different views. We have already touched on what they want to analyse and get out of that data.
Maybe I can bring in a personal anecdote. My son has had two MMR doses; one was in Spain, and the second one was in England. On his GP record, both doses are present, and I know that they are both present in the CHIS because I have looked. However, there is only one dose in the national dataset. I have even sat down with the practice manager to try to understand why this is the case. It highlights the issue when systems are designed around payment: a vaccine that has not been given here may be coded in a way that is not extractable and not visible to the systems. It might be about the way in which the GP’s IT systems are pulling that data, but we are undercounting to certain degree.
That is important to all the systems because, obviously, substantial resource is employed to try to reach unvaccinated children. When there is an outbreak, you are trying to contact parents and general practice to sort things out. The more accurate and complete the data is, the better it is all round. Everyone has a vested interest in improving it but it is complex and not always straightforward, although progress has been made.
Baroness Wyld: That is the question, I guess: everybody has a vested interest, but who is actually in charge of it? Who is going to make sure that things get better?
Dr Colin Campbell: Each organisation is doing this. There are national structures for managing the vaccine programmes, through which reporting has to be done, but the data are coming from different places and are being pulled through different systems to try to inform the national picture. For example, NHS England’s digital strategy of looking forward to a patient record or unifying that view of vaccination history is important, but it depends on having the accurate data behind it; that falls under different people’s responsibilities, I think.
Baroness Wyld: Do you think that the different parts of the system are talking to each other enough or effectively?
Dr Colin Campbell: From my own perspective, we meet with NHS England fortnightly and we talk to our colleagues in CHIS quite frequently, but some things are separate. For example, a GP IT system is commissioned separately, and we have no control over what happens there. We can produce a specification for it to implement but, when GP IT systems modernise their own systems, that can have consequences which need to be monitored, evaluated and fed back when it produces, say, changes to the data.
Paul Midson: On who is accountable for improving data systems, I think that, to some extent, everybody in the health system is responsible. We speak with our system suppliers and tell them, “This is what we need to be able to run our service effectively”. If we did not do that, we would just be stuck with the same system.
One of the problems, in child health terms, is the specification for a child health system. The information requirements specification was published in 2015 and has not been updated since. Obviously, the technology requirements of a child health service have changed dramatically in the past decade, but that specification is still the same. So the changes that are introduced by system suppliers for child health are driven by the child health teams themselves saying, “This is what we need”, or a new vaccine will come into play and they will need to do that via an information standard notice—or a new whatever it is that comes in. One of the main weaknesses of the whole system is the lack of a solid specification and an organisation that will hold the system suppliers to account. We are quite lucky; we have some good system suppliers, but there needs to be an overarching accountable organisation.
Baroness Cass: Before I come on to my question, is anyone, internationally, doing this better? The Nordic countries are normally good at it.
Dr Colin Campbell: Yes. Smaller populations are easier to deal with. Denmark, I think, has a unified system. Spain uses, for example, a municipal register as the basis for administrative data, which determines not just your vaccine status but your interactions with the local government system. That means that the denominators are probably more accurate than ours. As we have mentioned, we have a particularly mobile population, which is quite hard to deal with—people coming in and out of the country. We have done some work around that. We know that in local authorities where there is a lot of movement in or out, year on year, the vaccine coverage is lower. That is partly about capturing vaccine events through examples like mine, but it is also the denominator problem: are we accurately capturing that population movement in the data systems?
Q146 Baroness Cass: To get on to my proper question, we have talked about vertically sucking data, or pushing it uphill, from primary care records to CHIS and to COVER. Thinking about it more horizontally, and about how health systems talk to each other so that you can share information about a patient’s vaccination status, what are the main barriers and what is the best way to overcome them?
Simon Hailstone: We are getting into the topic of IG and data sharing, and in imms it is quite a complex space. Paul might want to expand on this but, certainly in London, data controllership for CHIS includes essentially all the GPs in the region. Making changes to the data-sharing agreement that is currently in place is a rather large undertaking, and it means that we sometimes do not have the flexibility to be reactive where it is justified. That is important in making sure that we are following the letter of the law and so forth, but it means that we do not always have the agency to make some of those decisions quickly about what we are sharing, although we can do a fair bit under the existing set-up.
I get somewhat more worked up about the interdepartmental data-sharing questions, which we have touched on. It came up in the previous panel, regarding the Department for Education and its role in this. Two elements are a particular issue for us in London. One is the sharing of parent contact details. Each school has its own privacy notice and its own interpretation of it, and it becomes really challenging to consistently try to bring all those schools on board with the idea that they can share parent contact details. Having a firmer stance on that would be quite helpful.
We also have issues around getting hold of things such as the school census, which is submitted to local authorities. That is what allows our CHIS to link children on its record to the school that they attend. Independent schools are not included in that census, nor are home-schooled children. Even the children who are in the census do not come with a pre-packaged NHS number next to their name. Our CHIS systems, sometimes using an algorithmic approach and quite often with a bit of manual heavy lifting, have to do the work to align and link those two data sources. That certainly plays its part, but IG and data sharing are complex in this space because of the number of actors at play.
Baroness Cass: Having listened to the previous panel, if I were on Mavis, in a school, I could access parts of the Spine. So could I, for example, tell if a child had had their under-fives tetanus vaccine?
Simon Hailstone: I am not certain on Mavis, off the top of my head, but we could follow up.
Paul Midson: We have supported some of our school-age vaccination services that have started to use the Mavis system. We are basically supplying them with the cohort list, the patients list and the vaccination histories, so those get fed into Mavis.
Baroness Cass: So that is a manual thing again. Paul, because CHIS is core in this, how well do you think the data system supports call, recall and consent for vaccination?
Paul Midson: As I mentioned earlier, the CHIS systems are highly configurable and they support it well, but the wider processes behind it might need some overhaul. For example, a couple of years ago we took on a contract delivering CHIS in the east Midlands. We looked at the core recall processes there. Only two of the five regions actually scheduled children for vaccinations with CHIS; the rest of it was done by GP practices.
In the areas where we schedule, it was based on an old method of specific dates and times. We would run a schedule every week to say which children were eligible for a vaccine, and write to the parents, saying, “You have been booked an appointment on this date at this particular time”, which did not give parents much flexibility. In order for that to work, the practice then had to have a set limit of appointments it could offer to fit all those children in. We have now changed that to a different method whereby we send a communication to the parents saying, “Your child’s eligible. Please phone your GP to make an appointment”. In doing that, within nine months, we increased the uptake by nearly 20% in over 80% of GP practices.
Dr Colin Campbell: We make efforts to support on local outbreaks, for example. Traditionally, when you are looking at vaccination status for measles, the health protection team has to phone the GP to find the vaccine status as part of the risk assessment for that case. We have tried to support that by providing the health protection team with access to the record-level data in the context of an outbreak but, of course, that depends on the data quality itself. It is fine if a vaccine record is present in that dataset but, because of the lack of completeness, particularly in older age groups, for which there is no vaccine record, the health protection team is still having to phone, and that obviously has resource implications. It is not ideal.
If I may piggyback on what Simon said about governance, during Covid, we had access to individual level ethnicity data, which included Gypsy, Roma and Traveller, for example, until August 2023. When the Covid regulations expired, that dataset no longer flowed to us. That, I think, is governed by primary care, and it has been really difficult to restart; we have not got that again. That is primarily affecting children born after August 2023. So in a measles outbreak, if we are looking for ethnicity in the new children, it is difficult to find. That is an example of some of the problems.
Likewise, on governance, we have wanted to get access to record-level national CHIS data to do things, particularly around safety, but we have also done some work around supporting the BCG programme in the change from SCID, and in the introduction of SCID screening. But London CHIS data is not available to us, and that is primarily governance driven.
Q147 The Chair: You have described a system that is multilayered and very complex, but for some families the problems are really quite simple. I will give you two examples and then ask you something about them. One is the family that moved house and had great difficulty in transferring its information from one GP to another, presumably because they use different systems. Secondly, we had in front of us a paediatrician who worked in hospital clinics, who said that, until recently, when her hospital trust changed its IT system, she was unable to get the record of having given a child in her clinic a vaccination on to the GP’s record. Is there anybody whose job it is to sort out things like that?
Dr Colin Campbell: I may have to defer to colleagues. We do not run the system, for example, in NHS trusts.
The Chair: Those things cannot be unique. There are probably loads of families who have trouble with systems when they move house.
Dr Colin Campbell: We have had anecdotes as well of people changing practice and the records not transferring, absolutely.
The Chair: Is there any system, person or body whose job it is to dig into these real, practical problems and sort them out, or at least identify what can be done to sort them out?
Simon Hailstone: It possibly gets to the level of escalation that that issue would get to, and whether the root cause is found to be an issue with an individual practice, in how the record was transferred, or whether it transpires to be a national system issue, which would dictate where it sits in the response and with the accountability.
The Chair: How many different systems are there that GPs and NHS hospital trusts use, and how many of them will not talk to each other?
Simon Hailstone: “Talk to each other” is a very wide term. How many GP systems are there?
Dr Colin Campbell: There are two main GP IT systems and one new one which is coming online.
The Chair: What about the hospital trusts?
Dr Colin Campbell: I cannot answer that. I am sorry.
The Chair: A very large number, I would say.
Baroness Neuberger: A large number, yes—two main ones, but quite a large number.
The Chair: It sounds as if that is a problem.
Simon Hailstone: I agree. This is where things like the shared record view could stand to improve things. Again, it does not open up that question of the right bit of read-write access to the record, and we are built on this model where—I do not think it is quite correct in how it is worded—the primary care record is the primary record of the vaccination. Therefore, we have this system which relies on shuffling records around the system, and that does introduce, in my view, a bit of lag to the system and more opportunities for things to go wrong.
Baroness Hodgson of Abinger: Following on from this, what level of confidence do you feel is desirable that you have got it all right? It sounds to me that to get 100% accuracy would be very difficult. To feel that you were on top of this, what would you desire?
Simon Hailstone: There are some unknowns in the system that we are still not in a position to quantify really accurately. Colin mentioned the technical term “ascertainment”: are we capturing all the vaccination information that is out there? The other question is: are we accurately measuring the denominator? The latter is really difficult to predict, because in effect, you are trying to prove a negative. How many people who do not pick up the phone are not picking up the phone because they are not responding, or is it because they are not there? There are multiple different causes of that. We recognise that there are genuine access issues, genuine problems with hesitancy and so forth in the population, but distinguishing those from these wider systematic issues is really quite challenging.
Baroness Hodgson of Abinger: Does anybody else have a view?
Dr Colin Campbell: Yes. Accuracy in the clinical record for the GP, for instance, is important because they are dealing with the person in front of them. National systems, where you are looking broadly at numbers, have a tolerance, and as long as we understand in what direction and roughly by how much things are wrong, we can produce good data that makes meaningful interpretation possible, so that is okay. But what we are describing, I think, is some uncertainty around how big that problem is in the national systems, and particularly around the denominators.
The Chair: We had better move on, I think.
Q148 Baroness Nye: Our exam question, if you like, is to look at the decline in childhood vaccination rates, and to do that we need to get to grips with how well data systems support national, regional and local surveillance of vaccination uptake, coverage and disparities. I apologise: I think you answered this question in previous answers. In your view, what are the main barriers to improving that surveillance and what would be the best way to overcome them?
Dr Colin Campbell: We covered quite a lot of that in answer to earlier questions about how we are trying to support local and regional systems. Of course, the data is trying to inform national policy based on modelling of current susceptibility, and that is particularly where we might have problems with older data being less accurate.
Baroness Nye: Is there anything, in your view, that slows down the targeted action? Obviously, you need the data to give you information that you can use to target action. Is there anything that slows that down, in your opinion?
Dr Colin Campbell: For the most part, when there is an incident, we are responding really quickly. When there is a safety concern, that is investigated immediately and we can pursue the data systems. It is more the complexities and the structural interrelationships that are difficult. On the point about the lack of ethnicity data, you can get around things like that—for example, by talking to the GP—but that adds another step.
Baroness Nye: Would you say that the data on vaccination uptake is very well shared with local public health teams, or could that be improved?
Dr Colin Campbell: Everything can always be improved.
Baroness Nye: In that case, how would you improve it?
Dr Colin Campbell: We undertake regular surveys of the data—for example, of the official statistics—asking what people want to see. We have held, or have started to hold, a series of workshops recently, asking people what they need. One of the consistent things is that we want more local data—for example, local authority level data. As a result of that, in the RSV programme, we actually added that into the backing tables, which is an improvement and a response to a specific request. We try, for example, in the publication of stats, to produce breakdowns such as local authority, ICB and, where appropriate, general practice-level data that can be used by local teams. But there is also, as Paul mentioned earlier, local access to data through the CHIS directly. For the most part, people have access, maybe not always as frequently as they would like.
Paul Midson: I agree. In terms of sharing data with public health teams, again, it really comes down to the variance in the CHIS service. Some teams might do that; some others might not.
As I mentioned before, we have developed these extra reports that we publish and share with NHS England and local public health teams, with different demographic influences in there as well, to give a more rounded view of uptake. Personally, I do not think we need to do anything to improve our process, but I am sure there are other processes that could be improved.
Simon Hailstone: In London, we are working on improving that. We have only recently started getting dashboard-type tools built on top of CHIS. This adds to the picture that how things are enacted across the country is inconsistent. Whether something such as that needs to be built more explicitly into the specification for CHIS, for instance, is a very valid question. It is even worth considering whether the underlying data structure that would inform a dashboard be made consistent across the country.
We are doing a lot more work recently on trying to build an end-to-end audit process working with the patterns we see in GP practices within COVER data and using that to inform targeted audits of what is going on in the records. For instance, if we have 19 unvaccinated or undervaccinated children in Westminster for Q3, we go back to the practices where those children are registered and work through it to understand: is it data quality? Is it waiting for evidence of international vaccination? Is it something else? One of those “something else” issues that we have certainly encountered is that practices have, in many cases, a number of children on their books whom they believe are inactive but they are not able to remove for legal reasons.
Ultimately, this is a policy issue in that things such as FP69—which is a term you may have come across; it is the national list cleansing process—applies to adults but not to children. That is for very sensible public health reasons, but it comes with the upshot that a GP’s child list will be inflated. If they do not know where children have moved on to, they will not remove them from their records. We have certainly come across that as a fairly new bit of learning for us in some of the practices that we see as outliers in the coverage statistics.
The Chair: In a very large ICB such as London, is it possible to get down to GP practice-level information?
Simon Hailstone: Absolutely, yes. If we want to, we can get lower than that with CHIS. Obviously, you start hitting more and more IG questions about what it is appropriate for different people in different organisations to access, but if we need to we can get to the single individual-level of detail working with CHIS, and we have done that on many occasions.
The Chair: Now, we come to the final question from Lord Dholakia.
Q149 Lord Dholakia: One of our aims is to make recommendations to the Government on how to improve data information, which ultimately assists greater use of child vaccination and more disparity. If you were to give your recommendations, what would be the two major recommendations that each of you would make?
Dr Colin Campbell: It is things we have heard before. One is to ensure that we are accurately ascertaining the numerator, so that vaccines that are given are recorded and coded appropriately and can flow through the system to be read by those who need to read them for whatever the purpose might be—and ensuring that there is a systematic approach to that, for example in vaccines given abroad, so that the records are consistent between systems. That is one: the numerator.
The second is the denominator. We need to ensure that denominators are clean and current, and that even if it is not appropriate to remove people from the denominator we can at least flag them, and that that flag can flow around the system so that we can deal with it appropriately when we see that flag, particularly when people leave the country.
Attached to the first recommendation—the numerator—is ensuring that there are appropriate places in contact with the health system to ensure that that cleaning can be done. GP registration, changing general practice and school entry are all areas where we could, if appropriately resourced and if we had the time, intervene.
Paul Midson: I have four written down. I will pick my top two. The introduction of automated mechanisms of transferring data to the stakeholders that need that information would be a great step in the right direction. Get rid of sending data via email. We need a standard transfer mechanism regardless of what system CHIS uses and what system it comes from.
Lastly, one thing I did not mention previously is consultation and collaboration. When these new systems are being developed, there has to be some sort of consultation or collaboration with the key stakeholders behind it, so that we reduce the wasted opportunities. We have seen it in the past when systems have been developed but are not quite how we how we would like to see them. There have been elements of that data transfer to CHIS that have been missed out, so it has been a missed opportunity because CHIS was not consulted. That would be my second recommendation.
Simon Hailstone: Mine would be a slight adaptation of Colin’s recommendation on numerators and denominators. We need to task the system to do what we want it to do, and to be very clear about that, and then things such as the resource and money can start to flow in that direction to fill some of the gaps that we have discussed here.
My other recommendation would be to reiterate those questions around how we can work with other departments—the Home Office and the Department for Education—so that the NHS programmes are furnished with the resources that they need to ensure that they have an accurate denominator, that we know on a countrywide basis which school a child is attending, and that we make those processes slick, effective and efficient.
The Chair: Thank you very much. That is very clear. Colin, Paul and Simon, thank you all very much indeed for all the information you have given us. If there is anything further that you think would be helpful to us, please do not hesitate to send it on.
