Automating Referrals with Blue Prism

Processing Ambulance referrals and entering them into Adult Social Care’s CMS (IAS) 

As part of the early-warning systems for any concerns that Ambulance staff have over the longer-term health and wellbeing needs of the patients they interact with, referrals are sent via the North West Ambulance Service (NWAS) via their online portal Cleric to local authorities to register the nature and the extent of these concerns. This automation – released a week ago – handles the data within these referrals and enters them onto Adult Social Care’s Case Management System, IAS, for a member of the team to parse over before confirming.  

Understanding the Challenge 

Referrals from NWAS come in the form of PDF files which are frequently around 5 pages in length. The data from these referrals must be read, understood and conveyed into the appropriate fields within IAS by a member of the Single Point of Access (SPA) team on behalf of Adult Social Care teams. Due to the sensitive nature of the contents of these referrals, these must be checked regularly and thoroughly to ensure that all relevant data is entered appropriately into IAS. It was estimated prior to the automation being developed that there were around 20 of these per week and that each referral could take between 15 and 30 minutes, depending on the complexity. Staff were notified of the receipt of a referral – which could occur at any time of the day, including weekends – via a notification email received to their shared team mailbox and assigned to an appropriate team member via Liberty Converse. 

As with any automated solution, an awareness of the risks posed was paramount throughout the design process. All data to be entered by the Virtual Worker is to be reviewed by a trained Single Point of Access (SPA) Officer before any decisions or actions are taken, ensuring that professional oversight – or ‘human-in-the-loop’ – remains central to the process. The process underwent rigorous testing with sample referrals and was piloted with oversight from SPA officers to ensure accuracy. Any anomalies are flagged for manual review, and a full audit trail is maintained. All data is handled in accordance with the council’s data protection policies, with secure access controls and audit logging in place to ensure compliance with GDPR. 

The Proposed Solution 

Utilising our in-house Robotic Process Automation (RPA) tool ‘Blue Prism’, it was proposed that a Virtual Worker (VW) would scrape the mailbox at four-hour intervals throughout the week to collect any new referral ID’s to process. From this trigger, the process would log in to Cleric, download the relevant referral PDF and scrape the data from it, storing this on the SPA file share for data security purposes. From this, the data would be formatted appropriately and entered into IAS. A more thorough breakdown of the process follows: 

Checking the Mailbox 

For the purposes of this automation, it was decided that, due to the number of other emails received to the shared mailbox, that a new mailbox would be created for the purposes of this automation and referrals sent from Cleric would be directed to this mailbox instead. Once an email has been received, it will sit in the mailbox until the automated worker performs a scrape (scheduled for every four hours from 2am onwards). This email, if valid, will have the referral ID pulled from the subject line and added to an initial work queue. 

Summary 

Downloading Referrals 

The process then fires up a browser window and logs in to the Cleric portal, where all patient referral data is stored. Login credentials are entered and then all valid referrals are pulled in to the process to be checked against the list of referrals the VW has received from its previous scrape. Once a match has been identified, the PDF containing the referral data is downloaded and stored appropriately within the SPA team file share. The referral ID forms the basis of the ‘Queue Key’ for identification purposes of this referral through the process, alongside the file location of the downloaded referral and other key information pertaining to the referral’s nature (whether it is a Safeguarding referral or an Early Help referral, and whether the patient has been conveyed to hospital). 

Additionally, referrals can receive updates up to the point where they are completed. To mitigate this, referral ID’s scraped from emails are checked against items already in Blue Prism’s queue. If an item is present, depending on its state within that queue, either the process will continue on as intended (if the item has not yet been entered into IAS) or it will trigger an email requesting manual entry (if an item has already been added to IAS by the Virtual Worker but it has not been accepted on Cleric by the SPA team member assigned to the case).  

Summary 

Preparing the Data for Entry and Locating the individual on IAS 

From the queue item, the data pertaining to the relevant referral ID is pulled in to the process and arranged in a manner suitable for entry into IAS. This includes basic validation checks (e.g. Date of Birth is present) as well as more advanced formatting techniques to ensure that the data pulled in is laid out in exactly the way that the SPA officers and Adult Social Care team require for smooth processing of the referral data.  

Once prepared, and with IAS also successfully logged in to using the appropriate securely-stored credentials, the process of identifying the service user on IAS can begin (as well as checking that their status is appropriate for data entry – i.e. whether they already have any active engagements registered and whether they already have any active safeguarding investigations being performed). This is performed preferably via their NHS Number as the unique identifier (which is frequently present on Cleric referrals, but not always so on IAS), but failing that, based on their surname, date of birth and post code.  

If the process can successfully identify an individual based on this criteria, the contact can be added to that service user’s record. However, if not, the process can also create a new service user record using the details provided on the referral. There are a number of data quality checks prior to this to ensure that this is appropriate, such as whether they have been conveyed to hospital (if it is not a Safeguarding referral, this responsibility falls on the part of the staff at the hospital), and if there are any active involvements currently logged against this user. 

Summary 

Check and Amend details on Service User’s Record 

To ensure that IAS contains the most up-to-date information relating to an individual, and with the referee from Cleric being considered a competent referrer, the data contained within the referral is deemed to be preferential compared to the existing data on IAS. A series of checks are carried out on the data contained within Cleric to compare them with what is held on IAS. These include Name, Address, Contact Numbers, Living Alone status, GP’s surgery and Ethnicity. For some of these (Name and Address), due to the possibility of typos or additional punctuation contained within the referral, instead of changing these directly, these are flagged as being different (if appropriate) in the form of a notification on the email to the SPA team, alongside any other pertinent information relating to the referral (for example, if the individual being referred is also the main carer for someone else).  

Summary 

Add Contact Record 

Once all personally identifiable information has been checked or amended as appropriate, the contact record can be created. Prior to this, the VW checks whether it has any outstanding ‘partial contacts’ linked to that individual. These could’ve occurred if there was a previous technical issue (such as IAS downtime) that prevented the VW from completing a contact on a previous attempt. If it does identify one of these, it will close it, and notify the IAS System Support team that a partial contact – now cancelled – required deletion, with the relevant information required to be identified by the team.  

The relevant data from the PDF is then entered into the relevant fields on the Contact record. Included within this are a number of potential ‘Text Area’ fields where data is formatted (entered with bold or underlined formatting applied, with appropriate spacing) in order to aid readability from the end-user of this data. Depending on the nature of the referral, additional tabs on the contact record may be completed also, such as the Privacy tab, the Safeguarding Tab and the Existing Relationships tab. Once the VW has completed its data entry task, the contact is reassigned to a team folder so that it can be picked up by another member of the SPA team to review before completing it.  

Summary 

Notify SPA Team of Work Performed 

There are a number of flags, or notifications, that could have been raised throughout the parsing of the data from the referral. These are pertinent facts that need to be highlighted to the SPA officer to ensure that the correct process is followed promptly and accurately. These are entered into an email send to the shared SPA mailbox, alongside links to pre-formatted documents containing the content from the referral (if it was not able to be entered by the VW for a technical or valid business reason) and the Cleric portal where the referral would have to be accepted again by the SPA officer. Due to the Liberty Converse mechanism in place on the SPA team mailbox, this email, once received, is allocated to a single available SPA officer with the appropriate ‘skill’ (i.e. access and ability to enter the data). They then complete the referral as they would, with all the data presented to them already in-place and without the need to enter it manually. 

Summary 

Benefits of Automation 

Although it is still early in the life-cycle of this automation, there have been a number of key benefits noted by the team: 

Conclusion and Feedback 

By leveraging RPA in this process, staff have been able to focus more on complex cases and offer timelier and directed support for residents. In doing so, elements within a referral that may have been missed have also been brought to the immediate attention of staff in a timely manner in order to aid outcomes for the residents of Cumberland. With the benefits currently being realised, and with the potential for additional automations within the SPA service area, the potential for RPA (with full awareness of the risks posed by any automated solution) to return an increasing amount of time to the highly-pressurised SPA team is continuing to grow. Here are some quotes from our recent Post-Deployment Review session: 

By John Hammond