SPARRA
         Peter Martin (ISD)
          Mandy Andrew
(Long Term Conditions Collaborative)
SPARRA

• What does it do?
   – Risk factors / Patient Examples
• History of Development
• How is it being used?
   – The Long Term Conditions Collaborative
• Current & Future Development
SPARRA

  Scottish
  Patients
  At
  Risk of
  Readmission and
  Admission

SPARRA is an algorithm for predicting a patient’s risk
of emergency inpatient admission in a particular year
SPARRA – Current Risk factors
  Age
  Gender
  Deprivation Level of Residence
  Number of previous emergency admissions
  Time since last emergency admission                    Inpatient/Day
                                                         Case history in 3
  Total bed days accumulated in the 3 years
                                                         years prior to the
  Principal diagnosis (last emergency admission)         risk year
  Co-morbidity – number of diagnostic groups
  Number of Elective admissions
  Emergency Admission rate (standardised) of patient’s GP practice


                                                    Historic Period
                                                   2007 2008 2009 2010

                                                       Predictor      Outcome

                                                       variables       year
Example: individual with very high
predicted probability of admission
• Predicted probability of admission   86%
• Male aged 67
• Less than one month since most recent admission
• 6 previous emergency admissions
• Glasgow – most deprived decile
• Most recent admission diagnosis: COPD
Example: individual with very low
probability of admission
• Probability of admission   8%
• Male aged 67
• 2 years since most recent admission
• 1 previous emergency admissions
• Lothian – 2nd least deprived decile
• Most recent admission diagnosis: Injury
Development History
2006
•   Initial Focus on those aged 65+
•   Base-data
     – Sources from National Inpatient/Day Case Data (SMR01)
     – Patients with >=1 emergency admission 2001-2003 (200K+)
     – Risk of admission 2004 – outcome was known
     – Deaths before end of 2003 excluded
•   Algorithm developed using multiple logistic regression
2008
Extension to those under 65
•   Modelling work repeated on an ‘all ages’ cohort (700K+)
•   Identifies 2 x high risk (50%) patients
•   Adopted within the SPARRA service January 2009
SPARRA MH – risk of psychiatric inpatient admission
SPARRA the ISD service
• Risk Scores generated quarterly for all relevant patients
   – >700K (previously 200K)
• Data relating to their ‘at risk’ population distributed to Health
  Boards, CHPs & practices
   – Chosen risk thresholds (often >50%)
   – Patient-level data for medium to high risk patients
          ID information
          Risks scores & factor values
          LTCs evident from inpatient/day case history
          Admissions related to substance misuse
           (alcohol/drugs)
SPARRA – current coverage




                     Very                   Acute
                                     ency
                                 merg ns    sector
                     High
                               E
                          SPARRA issio
                               adm
                   High risk
                          coverage


                  Medium risk



                  Lower risk




                                                     4
SPARRA – Current Development Strategy


• Priority is more comprehensive case-finding tool
   “Enhance SPARRA by expanding the cohort for whom a risk can be
    estimated beyond those with a recent history of hospital admission”
                              Scottish Government – National LTC Action Plan

• Need to look at other data sources that
   –will extend the cohort
   –contain risk factors that will improve discriminatory power
   – are likely to be available in most localities

                                A& E
                  Hospit al                  Communit y
                               NHS24                            Social Care   Prescribing
                 Admissions                   Syst ems
                              Ambulance


                                            Primary Care
                                          (General Pract ice)
SPARRA – Current Projects
• Using external data sources e.g.

    – Data held by Primary Care Clinical Informatics Unit, Aberdeen
     University on 40 practices and linked with national hospital
     admission data
• Maximising/simplifying use of hospital admission data
    – Admissions related to alcohol or drug misuse
    – Admissions related to falls
    – LTCs
• Streamlining our data generation/distribution process
   – Making using of ISD’s warehousing developments
   – Monthly updates
• SPARRA MH
   – Evaluating long-term role
   – Overlaps with ‘acute’ SPARRA cohort
Long Term Conditions Collaborative


‘Delivering sustainable improvements in
 person centred services for people with
           long term conditions’


      Improvement and Support Team
          Scottish Government
Policy Context
• Long Term Conditions Action Plan – June 2009
  – Person Centred Care & Mutuality
  – 7 Change Actions
• Linked to Long Term Conditions Collaborative
  High Impact Changes
• Integrating Policy Streams
LTC Collaborative Workstreams

                           Level 3
              Complex      Highly
              Case/Care    complex
             Management



              Specialist
                                Level 2
             (Condition)
             Management         High risk




                                       Level 1
                Self
             Management                 70-80% of
                                           pop
Complex Care Workstream
          Risk Prediction
             (SPARRA)
                                • Learning   Events
                                • Buddying
                                • Resources
                                • Whole Systems Improvement



Anticipatory               Care
Care Plans              Management
Current Developments & Practice

SPARRA                                Tests of Change
                                      •   East
•   Using external data sources
                                          – Workforce (NHS Forth Valley)
•   Maximising/simplifying use of     •   West
    hospital admission data
                                          – Care Management Model
•   Streamlining our data                   (NHS Lanarkshire)

    generation/distribution process   •   North

                                          – A3s and Anticipatory Care
•   SPARRA Mental Health
                                            Planning (NHS Grampian)
•   SPARRA development group              – Integrated Care Model (NHS
                                            Tayside)

Peter Martin & Mandy Andrew: SPARRA

  • 1.
    SPARRA Peter Martin (ISD) Mandy Andrew (Long Term Conditions Collaborative)
  • 2.
    SPARRA • What doesit do? – Risk factors / Patient Examples • History of Development • How is it being used? – The Long Term Conditions Collaborative • Current & Future Development
  • 3.
    SPARRA Scottish Patients At Risk of Readmission and Admission SPARRA is an algorithm for predicting a patient’s risk of emergency inpatient admission in a particular year
  • 4.
    SPARRA – CurrentRisk factors Age Gender Deprivation Level of Residence Number of previous emergency admissions Time since last emergency admission Inpatient/Day Case history in 3 Total bed days accumulated in the 3 years years prior to the Principal diagnosis (last emergency admission) risk year Co-morbidity – number of diagnostic groups Number of Elective admissions Emergency Admission rate (standardised) of patient’s GP practice Historic Period 2007 2008 2009 2010 Predictor Outcome variables year
  • 5.
    Example: individual withvery high predicted probability of admission • Predicted probability of admission 86% • Male aged 67 • Less than one month since most recent admission • 6 previous emergency admissions • Glasgow – most deprived decile • Most recent admission diagnosis: COPD
  • 6.
    Example: individual withvery low probability of admission • Probability of admission 8% • Male aged 67 • 2 years since most recent admission • 1 previous emergency admissions • Lothian – 2nd least deprived decile • Most recent admission diagnosis: Injury
  • 7.
    Development History 2006 • Initial Focus on those aged 65+ • Base-data – Sources from National Inpatient/Day Case Data (SMR01) – Patients with >=1 emergency admission 2001-2003 (200K+) – Risk of admission 2004 – outcome was known – Deaths before end of 2003 excluded • Algorithm developed using multiple logistic regression 2008 Extension to those under 65 • Modelling work repeated on an ‘all ages’ cohort (700K+) • Identifies 2 x high risk (50%) patients • Adopted within the SPARRA service January 2009 SPARRA MH – risk of psychiatric inpatient admission
  • 8.
    SPARRA the ISDservice • Risk Scores generated quarterly for all relevant patients – >700K (previously 200K) • Data relating to their ‘at risk’ population distributed to Health Boards, CHPs & practices – Chosen risk thresholds (often >50%) – Patient-level data for medium to high risk patients ID information Risks scores & factor values LTCs evident from inpatient/day case history Admissions related to substance misuse (alcohol/drugs)
  • 9.
    SPARRA – currentcoverage Very Acute ency merg ns sector High E SPARRA issio adm High risk coverage Medium risk Lower risk 4
  • 10.
    SPARRA – CurrentDevelopment Strategy • Priority is more comprehensive case-finding tool “Enhance SPARRA by expanding the cohort for whom a risk can be estimated beyond those with a recent history of hospital admission” Scottish Government – National LTC Action Plan • Need to look at other data sources that –will extend the cohort –contain risk factors that will improve discriminatory power – are likely to be available in most localities A& E Hospit al Communit y NHS24 Social Care Prescribing Admissions Syst ems Ambulance Primary Care (General Pract ice)
  • 11.
    SPARRA – CurrentProjects • Using external data sources e.g. – Data held by Primary Care Clinical Informatics Unit, Aberdeen University on 40 practices and linked with national hospital admission data • Maximising/simplifying use of hospital admission data – Admissions related to alcohol or drug misuse – Admissions related to falls – LTCs • Streamlining our data generation/distribution process – Making using of ISD’s warehousing developments – Monthly updates • SPARRA MH – Evaluating long-term role – Overlaps with ‘acute’ SPARRA cohort
  • 12.
    Long Term ConditionsCollaborative ‘Delivering sustainable improvements in person centred services for people with long term conditions’ Improvement and Support Team Scottish Government
  • 13.
    Policy Context • LongTerm Conditions Action Plan – June 2009 – Person Centred Care & Mutuality – 7 Change Actions • Linked to Long Term Conditions Collaborative High Impact Changes • Integrating Policy Streams
  • 14.
    LTC Collaborative Workstreams Level 3 Complex Highly Case/Care complex Management Specialist Level 2 (Condition) Management High risk Level 1 Self Management 70-80% of pop
  • 15.
    Complex Care Workstream Risk Prediction (SPARRA) • Learning Events • Buddying • Resources • Whole Systems Improvement Anticipatory Care Care Plans Management
  • 16.
    Current Developments &Practice SPARRA Tests of Change • East • Using external data sources – Workforce (NHS Forth Valley) • Maximising/simplifying use of • West hospital admission data – Care Management Model • Streamlining our data (NHS Lanarkshire) generation/distribution process • North – A3s and Anticipatory Care • SPARRA Mental Health Planning (NHS Grampian) • SPARRA development group – Integrated Care Model (NHS Tayside)