Criminal Justice Committee

Meeting date: Wednesday, March 9, 2022

Official Report 767KB pdf

---

Risk Assessment in the Justice System

The next item of business is consideration of the service and case management system that is used by social work and prison staff to assess risk management in the justice system. I thank the Cabinet Secretary for Justice and Veterans for remaining for this agenda item, and I welcome to the meeting Cat Dalrymple, who is deputy director for community justice at the Scottish Government. I refer members to paper 5.

In light of the time that is available, we will move straight to questions.

Cabinet secretary, I understand that one of your responsibilities is to consider first grant of temporary release applications for people who are serving life sentences. My question is a two-pronged one. First, how many of those have you had to consider in your tenure? Secondly, how many of those were based on an incorrect risk assessment?

I think that I mentioned in response to a similar question from Mr Findlay after my statement last week that I hoped to be able to give him a reassuring response. I have not granted any first grant of temporary release, but it is also true that that function has been carried out for a number of years now by the justice portfolio minister—in this case, that is the Minister for Community Safety. A reconciliation of the Scottish Prison Service system and the identified affected cases has been carried out, and eight cases in which first grant of temporary release have been granted were found. All of those have been looked at again, and I can confirm that no immediate or concerning public protection risks were highlighted. However, those eight cases will be further reviewed by the risk review group, which I mentioned in the statement last week, to provide further assurance.

Thank you. I presume that some prisoners will believe that they might have suffered a detriment by being kept in prison for longer than the risk assessment might have suggested that they should be. Are you anticipating any form of legal challenge from prisoners, given how litigious they can sometimes be?

If there were to be a legal challenge, it would be for others to take that decision. From my point of view and the point of view of the justice portfolio, the important thing is to ensure that we can identify whether that has been the case. That is why we are carrying out the risk review process that I have previously mentioned. It is quite a lengthy, detailed and technical process with different layers, but it will be very thorough. I do not know whether it is possible to hear from Cat Dalrymple about how that will be carried out.

I should mention an issue that might well come up. Last week, I mentioned that we had 285 open cases. We were concerned and, as of yesterday, we have been able to check every single one of those cases. Not one is giving rise to any public protection issues for us.

On other cases in which people might feel that they have been wrongly assessed, my understanding is that, with the different layers of checks that are carried out in the risk review process, that is unlikely to be the case. However, it might be a good idea to hear about that from Cat Dalrymple.

I will start with the first grant of temporary release. I think that a briefing paper was provided to members yesterday that provides significant detail about the approach that is taken to such decisions. I hope that it explains to members that applications for a first grant of temporary release are holistic decisions for the whole risk management team. It considers a wide range of factors and available assessments, including the outcome of a level of service case management inventory, or LS/CMI. It gives an indication across the whole risk management approach of the manageability of the risk within the community.

As the cabinet secretary has indicated, 285 live cases were identified in relation to the risk score and level issue. All those have now been checked—the last two were confirmed this morning. Social workers have been absolutely amazing in coming back and checking all those cases, and they are confident that there are no live, immediate or concerning public protection risks in those cases.

On wider assurance in relation to the first grant of temporary release cases, those cases will be looked at, because they are part of the 1,317 cases that are being passed on to the risk review group, which will be chaired by the Risk Management Authority. That group met for the first time yesterday, and it is agreeing terms of reference. A number of organisations—11 different types of organisations from throughout the justice sector, I think—are represented on it, and they will assess any wider impact in any of those cases.

If you think about the numbers, we are talking about 1 per cent in the whole system. That figure of 1,317 is a good sample size. When those cases are considered, I suspect that any impact will inform any future lessons and work that would need to be undertaken.

Thank you. I will not hog the questions; I am sure that everyone else wants to come in.

I thank the Government for its briefing, albeit it was not easy trying to digest 18 pages overnight, given the technicalities of the problem.

I want to get my head around the bigger picture. During last week’s statement, we did not have a lot of time to go into detail, due to pressures on chamber time. This is a great setting in which to do that. Is 1,317 the maximum number of cases that have had a wrong risk assessment as a result of information technology glitches, or could there be more cases and you need to do further work to find that out and how far back the problem goes?

We have been liaising with our IT provider on the second issue that the cabinet secretary highlighted. On the first issue, we know that those cases are the ones that are affected, and we have provided a high level of assurance around the open cases.

The initial advice from the IT provider has confirmed the second issue, which is about the risk scoring of alcohol or drug use, and it has indicated that that is likely to overlap with the initial issue. Work is on-going in that regard. That suggests that a very small number of cases will be affected, as the error appears to retain information following a score being revisited.

To address that, we are working with the risk review group and professionals in the system to identify what additional level of assurance we are able to provide on the existing cases in the system. That needs to be a balanced and proportionate approach because there is no suggestion that a significant number of cases are affected by the issue. We already have a high level of assurance on the cases that have been confirmed as being affected.

We are working with the professionals to work out what is the best way in which to provide that level of assurance with the existing cases in the system. We are taking advice from experts from the Risk Management Authority on what that will look like.

In relation to the on-going case management of individuals in the community, there are a number of different points at which a social worker will provide assurance around the LS/CMI assessment.

On the fact that a known issue has been identified in the system, we have been told by all the professionals that, at any point when a decision is coming up about an individual, such as a court decision, in relation to preparing a parole dossier or by the Prison Service around progression, they will assure that individual’s LS/CMI assessment and have confidence in that assessment. Although no decision is based on that assessment alone, it is part of the holistic approach. Then—

Sorry—I appreciate that there is a technical answer to a simple question, but the problem is that I have not heard the answer yet. I want this to be absolutely clear. There are 1,317 cases, of which 1,032 are closed and the rest are open. Is there the potential for other cases to be affected by the IT glitch?

My second question is linked to the first. If the issue goes back prior to the IT centralisation project—the cabinet secretary said in his statement that that might have brought the issue to light in the first place—surely that means that, for a number of years, the system was getting it wrong. What work is being done to identify how many other cases there might be in which risk was incorrectly identified? What do you think the scale of that might be? Are we talking about tens, hundreds or thousands of cases? How many prisoners have been released in the past 10 years? I suspect that that is a substantial number. Does the Government know how many people might have been wrongly risk assessed prior to release? I do not want to know just about current cases but about those going back 10 years.

13:00  

At this stage, there is no suggestion that a significant number of cases is affected. An end-to-end assurance process of that system, going through every different functionality part, requires to be done. I do not want to get too technical, but risk assessment is technical—the 18-page briefing paper probably demonstrates that.

The first part of the LS/CMI is a kind of screening tool that is used predominantly for court social work reports. That does not necessarily have a calculated risk score; there are eight questions, six of which are yes/no questions. There are a significant number of those on the system that we know, from the identified issue, are extremely unlikely to be affected. We cannot see how they could be affected at this stage.

Thereafter, once an individual is being managed and is on a community payback order, they are likely to go on to the more detailed LS/CMI assessment, which is where the risk score is tallied up and comes out with a risk level. That is only part of the detailed assessment. The secondary part is considered, structured, professional judgment, which is applied in relation to a number of different questions about that individual’s criminogenic needs, their pattern of offending and all their wider information. That professional judgment is applied within the secondary aspect of the risk assessment, so it is not just a score and a level. If that professional considers it necessary, they do the third part of the LS/CMI, which is a risk of serious harm assessment. We need to assure—through the IT—every part of each of the three different stages, and the on-going review of those types of cases will be user-led.

I will need to check the Official Report, but I think that the language that you used is that there is an expectation that the issue will not affect a large number of cases. However, the answer is that we do not know whether 1,300 is the absolute number of affected cases, or whether more people have previously been released because the system incorrectly scored them. We could find them if we delve into historical archives and look on a case-by-case basis, but that is a tremendous amount of work. When will we know exactly? When will that piece of work be finished, so that we have a much bigger picture?

To be clear, two pieces of work need to be on-going. The first is a piece of work under those 1,317 cases from the risk review group, and that will inform any future assurance that is required. There are a number of different outcomes from that group. The holistic nature of risk assessment—of which LS/CMI is one part—is likely to inform any future assurances with regard to other old cases in the system. We will take our advice from the experts; if they assess the cases and are confident that there has been no impact, that might provide a high level of assurance. It is not for me to sit here and say what would be appropriate, because I am not an expert on risk management. We need to make sure that we take proper advice on that.

We also have the wider review of open cases in the system, and that is very much about working with the IT provider, the professionals and the Risk Management Authority to identify what will provide that level of assurance and public confidence in those existing cases. I suspect that it is not likely that every case will be reviewed because, as an individual is being managed, there are automatic, built-in review points. Obviously, when a decision point comes up, we expect a professional to have a look at the LS/CMI and make sure that it is assured and accurate. Thereafter—as I understand it, from the professional social workers who I have spoken to—individuals are ordinarily reviewed every three months. Therefore, within three months, I like to think that a lot of the individuals among the open cases will have had their case reviewed.

The language that is being used is around live or known public protection issues, but is the cabinet secretary confident that no one has been released earlier than they should have been? If anyone has been released earlier than they would have been under normal circumstances—were it not for the IT glitch—did any of those people, at any point in the past, pose a public protection issue? Outwith normal reoffending rates, which we talked about earlier, did any of those people go on to reoffend or end up back in the system? I guess that we are looking for a little bit more comfort that those who were released inadvertently did not go on to reoffend.

I think that it is right that we wait for the further processes that Cat Dalrymple mentioned to be gone through before we can be absolutely definitive, but if it is comfort that is being sought, I would just highlight two things. First, the fact that not one of the 285 cases in question has given rise to public protection concerns in the eyes of the experts who have looked at them is a good indicator of where we are at. Again, that is not definitive, and I am not trying to pretend that it is.

Secondly, in the light of some of the publicity that flowed from last week’s statement, I point out—and I cannot remember whether this is in our briefing or another that I have seen—LS/CMI is almost a general triaging tool. For someone who is, say, a sexual offender or who is seen as high risk to the public, because of violence, other tools as well as LS/CMI will be involved. That should provide you with reassurance.

I know that there was no time to take my opening statement, convener, but perhaps I can make a couple of other points that I highlighted in it, because I sought to address some of the questions that members, quite rightly, asked me last week in the chamber. Pauline McNeill and, I think, Jamie Greene asked about the technical nature of the update. I hope that members will see from the briefing that has been provided and from what Cat Dalrymple has said how technical the issue is, although I should say that I specifically asked for the language in the briefing for members to be as plain as possible, because it is sometimes quite difficult to understand the different aspects.

Moreover, in response to Stephen Kerr’s question last week about when all of this first came to light, I said that it was 24 January. However, the first person to see it was apparently an SPS individual on 13 January. I might have said that, too, but I certainly mentioned the 24 January date. That was when Government officials became involved and started running tests in parallel with the system. As I said last week, I was advised of the matter the previous Friday and then came to Parliament.

With regard to how quickly we came to Parliament, I would highlight a case down south that related to 400,000 prisoner or offender records, and there was no statement to the Parliament down there until after the event. I was keen to ensure that we did not do the same thing, and we therefore came to the Scottish Parliament as soon as possible. Coming back to Jamie Greene’s question, I point out that one consequence of that decision is that we do not have all the answers, because we are still working through this. We are providing as much assurance as we can—indeed, I think that the assurance that the 285 live cases have not raised any cause for public concern is pretty substantial.

There are probably one or two other questions that were asked last week that I have tried to deal with if not in the briefing then in the opening statement that I would have made. I hope that that gives some reassurance, too.

I apologise, cabinet secretary—I was just very aware of the time, which is why we went straight to questions. I am sure that members will have further questions to ask.

Is that you, Jamie?

I am keen to let others in, convener. If anything jumps out at me, I will jump in again.

In that case, I call Katy Clark.

I want to pick up the point that was made about other systems. Have you been able to work out the profile of the cases that we are talking about, particularly some of the more serious offences? How many sexual offenders are involved, for example? Given the huge workload that is involved in reviewing the cases, how have you prioritised them? Have you been able to prioritise some of the cases that would cause the public most concern? Perhaps you can outline your approach in layperson’s language.

I will respond first, as I believe that you asked a similar question in the chamber last week, and what I said in my statement should provide you with some reassurance.

You have asked about the offences that are involved, particularly whether they are sexual ones. I am grateful for the question, but I have to point out that LS/CMI is not an offence-based system and does not record the offence involved. As I have explained, it is a kind of generic triage system; although that might seem strange, that is entirely consistent with the risk assessment approach. As our briefing paper sets out, it is a general tool that looks at general factors, including potential offending, but part of the judgment applied relates to a different part of the assessment that looks at the nature of offending and provides structured consideration of that issue. I know that we have said this a number of times, but it is important to get across the fact that every risk assessment has different elements.

That element of the system does not have a score. For example, individuals convicted of a sexual offence will have bespoke risk assessments carried out. Those will likely focus not only on general offending but on risk of harm, which are particular to that type of offending. So, there are other processes that cover that.

You asked about what has priority. So far, we have concentrated on—this has been our priority—the 285 live cases. Moving on—

Sure. That is why I asked whether you were able to use other systems that you have in place to pick out the types of cases that might be of greatest concern. I wondered whether that has been incorporated into the work that is currently going on, so that the cases that were dealt with first were those involving the most serious offenders or offenders who have committed the types of offences that people would be most likely to be concerned about. Are you able to do that with the systems that are available to you?

I would make an assumption that the professionals in the system know the type of offending that is associated with the individual that they are managing. That is why we are working with the professionals to identify what level of assurance and type of review we do for any other open cases in the system. We will work with them and the Risk Management Authority to identify what those priorities should be. One of the messages that we have been getting from our social work colleagues is that they want clarity about what we are asking them to do. We will ensure that we provide such clarity.

Is it the case, therefore, that cases involving serious sexual offenders or individuals who have been involved in serious violence would be the kind that professionals would be asked to deal with first?

That certainly could be the case, but I think that the point that the cabinet secretary made is that looking at an index offence does not necessarily help you to understand the risk that an individual presents. You could have someone on a relatively low-tariff offence that creates a greater risk, and you could have the reverse.

Before coming to work in the Scottish Government, I spent 20 years as a procurator fiscal, and I remember that, in one case that I prosecuted on indictment, if you looked at the index offence, you could have assumed that there were five breaches of the peace, but the offence was clearly of a sinister nature. That case ended up being referred to the High Court, where assessors who were accredited by the Risk Management Authority produced a detailed risk assessment for an order for lifelong restriction to be considered. That demonstrates that the index offence is not necessarily linked. However, I understand the issue of public concern, and we can feed in that assurance level when the professionals tell us that that should be prioritised.

You are asking professionals to look at their current and former case loads and make a judgment about any individuals that they have concerns about, and for those cases to be prioritised.

We are certainly working with the professionals, but we have not asked them to do that yet, because we are working with them to identify the parameters, so that we can provide clarity on that.

You have not actually asked professionals to do that yet.

We met them on Monday to try to work out what the parameters of the review should be, and we will then take advice from the Risk Management Authority to ensure that it is content with our proposals.

Concerns were first brought to people’s attention on 13 February, although I appreciate that the full extent of the problem might not have been apparent at that point. However, as yet, we have not got to the point at which cases are being looked at by those who deal with this work. Is that right?

The priority has been the 285 live cases. Since we identified the cases, the professionals have been providing that assurance. The last two cases were reviewed overnight, and confirmation on those was provided this morning.

As I have said, the priority was the 285 live cases, which have now been completed. That is the priority that we set for those involved.

I should say that it was 13 January when concerns first came to light, and further tests were done by Government on 24 January, leading up to 25 February.

On the previous point, about the types of offences, those 285 live cases would have been subject to the LS/CMI, which is why we made them a priority. However, I again mention that the ones that involved sexual offences or serious violence would have been subject to a different tool that manages that high level of risk. What Cat Dalrymple is now talking about is going further and considering previous cases to ensure that we are satisfied with the rest of the system, too.

For the cases that have been looked at, have you asked for a breakdown of the offences that are involved in that cohort?

I will let Cat Dalrymple come in, but our priority was to make sure that there were no public protection issues. As has been mentioned, the index case for which somebody was first convicted sometimes does not give the full information.

I do not know whether Cat wants to add to—

13:15  

I appreciate that somebody might be a dangerous individual but might not have been convicted of any serious offences. However, many of that cohort would have had convictions. Did you ask for that information?

I do not recall that that information was asked for. The initial reassurance was sought as soon as the issue was identified, and the priority was checking that there were no concerning or immediate public protection risks.

I am watching the clock—

Sorry, convener, but I have a brief question.

Very quickly.

Convener, I have a question as well.

Apologies—I did not realise that you were waiting to come in, Fulton.

I will let Jamie Greene go first.

It is a supplementary question on a technical point. Am I correct in thinking that, if someone has been released, there is no recall to prison?

I also want to know whether there is anyone who is currently due for automatic early release rather than assessment-based release but who may have been incorrectly risk assessed. If so, will there be a moratorium on their automatic early release if there is the potential that they have been wrongly risk assessed by the IT system?

I think that you are talking about release from prison. Long-term prisoners who have sentences of more than four years are released after a Parole Board decision. That is set out in the briefing pack that we provided for the committee. Shorter-term prisoners who are serving sentences of under four years will be released at the statutory point in their sentence. As was highlighted earlier, there are statutory provisions on what throughcare they are obliged to take.

Would such people be released anyway due to the policy on release, even if they have been wrongly assessed? That is the crux of my question.

In relation to short-term prisoners, the release date will have been set as part of their sentencing in the first place. In relation to longer-term prisoners, it will be a decision for the Parole Board, which will have all the different experts to provide the risk assessment at that time.

Perhaps I can provide further reassurance. Everybody has now been made aware of the issue, so all current assessments are being looked at in the light of that. In any event, that will quickly be overtaken by the three-month assessment to which offenders in that situation will be subject. Therefore, even if we had discovered something, the process is starting to bite whereby such matters will be taken into account in future assessments.

Cat Dalrymple, as a former procurator fiscal, will correct me if I am wrong but, with short-term prisoners, the original sentence is handed down by the court, and we do not have the ability to change that.

I apologise again to Fulton MacGregor, who I will bring in now.

I apologise, too, convener. When I indicated earlier, I was not sure whether you thought that I wanted to come in or did not want to come in. It was also my fault.

I draw members’ attention to my entry in the register of interests, which shows that, before I became an MSP in 2016, this is exactly the sort of work that I did. I am well aware of the LS/CMI system, having used it several times a week or, more likely, several times a day. I did the initial training in 2012, when the system changed. For what it is worth, I think that it is a very good system. Therefore, I might be able to ask some helpful questions on it.

Cabinet secretary, you have spoken a wee bit about this, but do you accept that it is not just the LS/CMI system that is used and that there are a range of risk assessment tools? Perhaps to put members at ease, I point out that you would not just use the LS/CMI system and say, “Computer says yes.” Has that point been made to you when you have been speaking to people?

Yes. Last week, during the statement, I read out the different categories of professionals who consider cases. As you will know better than any of us, Mr MacGregor, a whole list of experts examine risk management, and the LS/CMI is one tool in which to do that.

Katy Clark queried whether we have asked for a breakdown of the offences. We have not. We could do that—and I will look into that—but the simple fact is that the individual social workers and other professionals involved know exactly what the offences are. They are the ones who apply professional judgment. The fact that they do that, with other experts to help them, is the most valuable part of the system.

I have made the point a number of times—although it seems to be lost sometimes—that it is important to realise that the LS/CMI is one part of the system and that other tools are used in addition to it when it comes to sexual and high-risk offending.

Have all the cases that you identified and have been working through been subject to the professional override that you talked about in the chamber last week? I will come back to that issue. Was the level of risk lowered from what the LS/CMI said in all of those cases? Do you have that information?

Are you asking about the clinical override?

Yes.

We have confirmed that the risk level has changed in both directions in those cases. The clinical override had already been applied in a number of those cases but not in all of them.

That is the point that I hoped to make. I felt that, last week, some of the questions implied that the situation was about the clinical override bringing down the risk to that in the LS/CMI. However, it is as likely, if not more likely, that, if the LS/CMI indicates low risk but professional judgment suggests higher risk, a higher risk is put in. Do you accept that point, cabinet secretary?

Last week, I gave the example of the issue that we have with the alcohol assessment, whereby the lower score that might be justified by somebody moving away from alcohol addiction might not have been captured.

The professional override—I say this as a layperson and I am sure that Cat Dalrymple will correct me if I am wrong—has now been applied again to 285 cases. The experts involved have looked at those 285 live cases and said that there are no public protection issues arising from them. As others have urged us to do, we now need to go back and look at previous cases. If 285 out of 285 cases have come back with no issues, that is a pretty good indicator of things. However, we accept that we must look at previous cases is in the interests of public reassurance.

One of the strengths of the literature on the LS/CMI tool in comparison to all the other risk assessment tools that are available is that professional judgment and the override facility are available.

I agree that the override function is an important part of the LS/CMI. Do the cabinet secretary and his officials accept that the process for using the override is robust? I do not want anybody to think that it is a case of an individual social worker or other worker applying an override and that is it. In most cases, the matter needs to go through several levels of management. The higher-up management will have a higher level of experience. You will have seen from the forms that are completed that a narrative around the justification for the decision in either direction is needed. Will the cabinet secretary join me in offering that reassurance?

I was aware of some of that but not as aware as you are of the detail. The system provides further reassurance. As you say, it is not the case that something is taken from one computer system and drives the whole process. Professional judgment should not be rushed past. It is an important part of the system, if not the most important part. As you say, that does not involve just one professional. The decision is checked again by others to ensure that there is nothing that should be questioned. It is a robust system.

Obviously, we regret the two issues that we have had with the IT system. We must and will learn from that. There will be an on-going process with the providers to ensure that we try to cover that in future and make the risk assessment system as robust as we can. It is and has been robust. The two issues with the IT system have given us concerns, which we have addressed, and I have said that I will come back to Parliament. However, we want to have the most robust system possible.

One feature of the Scottish justice system is that it has more people on remand and in prison than is the case in many other systems. Given that, we could be accused of being risk averse. However, we must have in place a proper system. We should be accountable for anything that has not worked as it should and for ensuring that we get things remedied as quickly as possible. That is what we are trying to do.

I hope that I have been able to use my experience on the matter to ask the cabinet secretary some questions that would be helpful in reassuring the public. I will now ask my most challenging question.

I return to the issue of training. There have been question marks over people’s confidence in using the override and their professional judgment. That is why it has that level of management experience around it that I mentioned. Having spoken to former colleagues, I am aware that it is likely that people’s confidence has been impacted by what has happened. What steps is the Government likely to take to support people in the profession to bounce back from the situation, feel confident and not end up having more work to do as they try to make risk assessment decisions? Are there funding and resource issues to consider, too? There will be a confidence issue now.

I briefly touched on the point about funding last week. We have made it clear that we are willing to respond to any request for financial support that derives from a need for more people and resources to look at the matter.

There are a number of levels to rebuilding people’s confidence. One relates to the IT system itself. We will engage with IT people and others to ensure that the system is as it should be. Beyond that, as a number of members have suggested, we need to go back to closed cases and ensure that the system operated as it should have done despite the IT issues. Those two things should help to provide confidence.

We are also taking the opportunity to see whether we can make further improvements to the system. We should do that at any point but this seems the right point at which to do that.

I have a brief comment rather than a question. Having listened to the evidence and read the briefing note, I feel reassured. My initial instinct was that the decisions are largely based on the judgments of a series of professionals and were not so much about a computer. That is what I felt from the start and it has been confirmed, so I am reassured.

In the spirit of timekeeping, we will bring the matter to a close. I apologise once again for my sloppy convenership in forgetting to bring you in, Fulton, and for not allowing you to make your opening statement, cabinet secretary. You are, of course, welcome to make any additional final comments now or share information with the committee in due course. Finally, I ask for reassurance that the committee will be kept updated on the on-going review.

I am happy to give that assurance. It is an on-going process, as you said, convener, so it might be worth providing more than one update as we go through it. However, we will make sure that the committee is kept informed.

That is perfect. Thank you, cabinet secretary, and Ms Dalrymple.

I bring the public part of the meeting to a close. We now move into private.

13:29 Meeting continued in private until 13:51.