Data Analysis as part of the Assessment Process

1
Data Analysis as part of the Assessment Process

• Jill Burke
• Insight Solutions

2
Who are we?

• Established since 2001
• Independent IT training consultancy
• Not affiliated to any of the clinical suppliers
• Designated Clinical System trainers for all major
  systems
• Working with 2500 practices
• Working in over 75 PCOs/LHBs
• Accredited training Company
• main contract software suppliers
• Welsh Assembly Government
• nGMS Contract Training specialists
• IT Training Specialists for Primary Care

3
Ethical Disclaimer

• All information provided during this presentation
  is an opinion and is, therefore, optional. It is
  designed to enhance your abilities to work with
  the nGMS Contract should you wish
• It is based on how QoF 2 stands today! There are
  likely to be many changes, it is the
  responsibility of the practice to ensure they
  keep up-to-date with any future changes made
• The decision to implement any changes rest
  entirely with each practice/clinician

4
Why?

• Assessments in the past have often been routine
• Assessors going through learning curve
• Practices expecting you to look at patient data
  for evidence
• Largest investment in Primary Care
• Practice income higher than ever before

5
Where are we now?

• QoF2
• Established a process of assessing
• Identified a need to move this forward
• 2 years of back data and information from QoF
• years of other information
• Prescribing data
• Prevalence
• 2 years experience of assessing
• But not necessarily using all the information
  together

6
Where do we want to be?

• Assessing practices
• Accurately, consistently and relevantly
• Justifying payments
• Confident practice have achieved as a result of
  hard work
• Not just ticking boxes

7
Benefits outside of QoF?

• Chronic Disease management
• Powerful information for commissioning of
  services
• Community Matrons
• Admissions Prevention
• Practice Based Commissioning
• Accurate Disease Prevalence
• Budget Setting
• Better Patient Services
• Patient Pathways

8
Data AnalysisA new dimension
9
Three Approaches

• Comparing expected and observed prevalence rates
• Identifying high levels of exception reporting
• Comparing prescribing and prevalence rates
• Approaches will identify lines of enquiry not
  definitive answer to questions of gaming
• Making assessments worthwhile

10
1. Prevalence Rates

• QMAS reports crude prevalence rates not
  adjusted for risk factors (age, sex ethnicity
  etc)
• Very high / low rates might be explained by these
  factors
• Doncaster PCT have generated a tool that refines
  expected prevalence rates for 7 conditions based
  on practice level socio-demographic factors

11
Data required

• Practice Level
• Reported prevalence rates (QMAS)
• Age sex breakdown
• Deprivation
• PCT will need to data entry into spreadsheet
  time consuming but worthwhile results

12
Model Output
13
(No Transcript)
14
(No Transcript)
15
2.Exception Reporting

• Exception reporting codes concerns
• over-use
• under use
• misuse
• Comparison can be made with reference to PCT
  average level of exception reporting and size of
  disease register
• QMAS provides data on levels of exception
  reporting

16
Automatically identifies indicators where the
practice varies from PCT average
17
Locate data in QMAS
18
Download into Spreadsheet
19
A few simple calculations
20
Produce Control Chart
21
(No Transcript)
22
3. Comparing Prescribing and Prevalence Rates

• One would expect strong association between
  prevalence rates and prescribing rates for
  chronic diseases
• Variance from expected might indicate inaccurate
  registers
• School of Medicines Management (Keele Uni) have
  investigated this approach

23
CHD Register Size vs Chapter 2 Prescribing
EXAMPLE
24
CHD Register Size vs Chapter 2 Prescribing
Poor Data Quality? Patients not identified on
Registers
Poor Data Quality Inflated registers?
25
Practice to benefit

• If prevalence is high and incorrect, practice
  targeted to deliver services to patients who not
  in need
• More work
• Less success
• Fewer points
• If prevalence is low and incorrect , practice may
  be doing the work but not getting recognition
• Less per point

26
Patient to Benefit

• Correct services offered to relevant patients
• Patient care to improve
• Practices dont waste patients time

27
PCT to Benefit

• Correct investment of public money
• Patient services improved
• Practice achievement improves

28
Other possible checks

• Hypertension prevalence vs anti-hypertensive
  prescribing
• BP5 vs anti-hypertensive large amount of points
• Asthma and COPD prevalence vs BNF Chapter 3
  prescribing
• Diabetes prevalence vs diabetic prescribing
• Epilepsy prevalence vs anti-epileptic prescribing
• CHD8 compared with lipid lowering prescribing

29
What to target

• With so many areas / indicators / disease
  registers be selective
• Identify which registers / indicators are best
  for target
• High points potentially expensive for PCTs if
  there is concern
• Speciality within practice/PCT
• Diabetes
• Asthma etc.

Use data analysis to establish this
30
High Point Indicators

• CHD5 BP less that 150/90 - 19
• CHD8 Cholesterol less than 5 - 17
• BP4 BP recorded in past 9 months - 20
• BP5 BP less than 150/90 - 57
• DM20 HbA1c 7.5 or less - 17
• MH9 review in past yr - 23
• Asthma3 variability/reversibility - 15
• Asthma6 review - 20
• DEM2 review - 15
• DEP2 assessment - 25
• AF3 anticoagulation/platelet - 15
• Smoking - 68

31
Best use of data

• Pointless to have all this data and use it in
  isolation
• Cross referencing all sources
• E.g. Prescribing versus prevalence
• Prescribing versus achievement
• Exception reporting versus Prevalence
• Practice to practice comparisons

32
Summary

• Data will need to be collected, prepared,
  analysed and understood before the assessment
• Assessors should have the questions ready to ask
• May need practice input for some data
• Standard approach across PCT for all practices
• Target areas pre defined
• Training if required
• Assessors need to see the benefit
• May require specific training e.g. Excel