Enhanced SPARRA Predictive Model SPARRA Patient Alerts

Enhanced SPARRA Predictive Model & SPARRA Patient Alerts Risk prediction and service development Kathleen Mc. Guire – Long Term Conditions Manager Ehealth LTC Workshop November 2011

Aim • Integrated systems and communications • Expand the cohort of patients for whom a risk score can be calculated over and above the current SPARRA “All Ages” (Version 2) algorithm • Improve the predictive power of the algorithm • Provide the board with a risk prediction tool which will identify patients for referral to Community Wards

Scope • Feasibility of a model which included GP/Primary Care, Social Care, Accident and Emergency and Prescribing data. Any other potentially useful sources which may serve as a predictor for emergency hospitalisation, including falls and Io. RN. • Producing a linked data set • Improved data links to and from Primary and Secondary Care

Deliverables • Data extract specification • Predictive risk model (with technical report and recommendations) • Reporting and implementation scoping requirements (report) • Prototype reporting tool (with user manual) • Final report (with conclusions and recommendations) • Post project evaluation (report )

• Lessons learned will inform the national development of SPARRA, predictive risk modelling, related tools • GP SCI Gateway referral message integrated into ADASTRA

How we have taken forward • Appointed a Data Analyst • Appointed a Project Manager • Gained strategic organisational commitment- e. Health programme • Gained buy in by utilising other national SPARRA developments & improvements • Integrated the project with service development of Community Hubs • Expansion of current LES

SPARRA Prediction & Tools Accident & Emergency s ervice Navigator TS LO LES SPARRA l. S Socia Information Tool ng rsi Nu mes Ho I SPARRA C O AD Patients/Carers

Community Hubs SECONDARY CARE S NH District Nurse AY R O SH SP IR IC E E S Practice Nurse HUB Social Care Liaison RVICE Enhanced SPARRA CPM GP S P O C AHPs Pharmacy OT Specialist Nurse Geriatric ANPs AL SE Multiple/Complex Social-needs Patient PRIMARY CARE TEAM Intermediate Care & Enablement Teams SOCI Chronically ill Patient 24 Acutely ill Patient Community Wards H C O AD CARERS (Kinship / Professional)

What we have learned so far • Project needs to be integrated with other development • Requires stakeholder buy in • Differing views around data specification • Quality and coding of data • Informed consent and data sharing

What we have learned so far • Time, expertise and partnership required • IT systems used in out of hours setting suffer from poor demographics and duplicates • Integrating Primary Care systems with the out of hours service requires a primary data cleansing task

Our wish • To use the principles of risk adjustment to evaluate the pathways of complex communitybased interventions to reduce avoidable hospitalisation, eg testing the cost effectiveness of Community Wards and Telehealth • To link large datasets at an individual level pseudonymously through our partnership arrangements and relationships • Predict future costs of health and social care

Our wish • To exploit new Clinical Portal technology to help distribute electronic information to the point of care. • Successful procurement of a new community wide IT system (currently in progress) to support collection of data and distribution.

Requirements & Next Steps • National Support and continuation • CHI Seeding, time & expertise • Integrated IT systems within and across organisations • Go live with model 1 st April • Extend roll out and testing of CPM • Further integrate SPARRA & IRF