Scoring error in the LS/CMI system: FOI release

Information requested

You asked for information regarding the scoring error in the LS/CMI system.
Please find a response to your requests below:

1) How long the error had been occurring for, before it was identified in January of this year?

2) How many prisoners were wrongly scored in total using this system?

3) How many prisoners had scores which deemed them higher risk than they should have been?

4) How many prisoners had scores which deemed them lower risk than they should have been?

5) How many prisoners were scored using the system and it is not yet known if their scores were accurate or not?

6) How many prisoners were released prior to the error being identified, based on scores from theLS/CMI system.
 - and of these:
a) Have any prisoners who were released prior to the error being identified had their cases reviewed?
b) Have any prisoners who were released prior to the error being identified, been recalled to prison?
c) were any prisoners who had been released prior to the error being identified found to have been scored as lower risk than they should have been?

Response

1) How long the error had been occurring for, before it was identified in January of this year?
A. Investigations are underway as to the cause and duration of the LSCMI risk scoring/level IT error. The Scottish Government are working closely with the LSCMI IT managed service provider and a third party IT provider to continue investigations and testing on the LSCMI IT system. Users have returned to the paper based assessments until that assurance work on the IT system is completed. A review led by the Risk Management Authority is also under way and will report their findings in due course. The Scottish Government have already concluded the review of the 285 open cases that were identified as possibly being affected by the risk scoring/level IT error and no public protection issues were found in any of them.

2) How many prisoners were wrongly scored in total using this system?
A. Initial work identified up to 1,317 assessments, 285 of which involve open cases, had the potential to be affected by the risk scoring/level IT error. On Tuesday 1 March 2022, Local Authorities and the Scottish Prison Service were asked to urgently review the 285 open cases, to report back any discrepancies found in the scores and displayed risk levels, and consider if there were any, whether those presented a public protection risk. The review concluded there was no risk to public protection.

3) How many prisoners had scores which deemed them higher risk than they should have been?
A. Analysis of the impact on any key decision points will be considered by the Risk Management Authority as part of their review of cases.

4) How many prisoners had scores which deemed them lower risk than they should have been?
A. Analysis of the impact on any key decision points will be considered by the Risk Management Authority as part of their review of cases.

5) How many prisoners were scored using the system and it is not yet known if their scores were accurate or not?
A. Analysis of the impact of any key decision points will be considered by the Risk Management Authority as part of their review of cases. In addition, social workers are currently undertaking a review of all open cases to identify any issues, are undertaking paper based re-assessments when required, and will continue to review assessments as part of managing offenders.

6) How many prisoners were released prior to the error being identified, based on scores from the LS/CMI system.
- and of these:
a) Have any prisoners who were released prior to the error being identified had their cases reviewed?
b) Have any prisoners who were released prior to the error being identified, been recalled to prison?
c) Were any prisoners who had been released prior to the error being identified found to have been
scored as lower risk than they should have been?
A. The ongoing review work will look at cases where the LSCMI assessment related to a prisoner. As part of the initial checks, the Scottish Prison Service (SPS) identified 8 First Grant of Temporary Release (FGTR) cases to be checked within the open and closed cases, and found no public protection issues with any of these. 7 of these individuals are in custody and one has been on licence since 2019. Before Scottish Ministers consider an FGTR application, the SPS undertakes a comprehensive assurance process, reviewing the application from the relevant multidisciplinary Risk Management Team (RMT) against a clear set of standards and guidelines to ensure decisions are accurately recorded and founded on appropriate evidence.

It is perhaps worth noting that there are processes for each decision point in the justice system that mean the use of the LS/CMI tool is always part of a broader assessment. For example, the Parole Board does not rely on this information alone and will consider a number of factors and reports when considering release, including (but not exclusively): the circumstances of the (main) index offence, and any offending history; other reports produced; an offender’s conduct since sentence and intentions if released; all relevant information in the dossier; and the evidence heard at the parole hearing. The LS/CMI system itself – and as detailed in its guidance – includes a standard function to override the risk level displayed, and this is routinely used in the exercise of professional judgment by social workers.

About FOI

The Scottish Government is committed to publishing all information released in response to Freedom of Information requests. View all FOI responses at http://www.gov.scot/foi-responses.