What is the Prevalence of CHD and Diabetes

1
What is the Prevalence of CHD and Diabetes?

• Assessing the completeness of GP registers using
  QOF data

11th June 2007
Jennifer Nicholas - Public Health Intelligence
Epidemiologist
2
Outline

• What is the QOF?
• Are chronic disease registers important?
• Chronic disease registers measuring prevalence?
• Estimated prevalence vs QOF disease register
• Conclusions

3
Quality and Outcomes Framework

• Component of GMS contract introduced 1st April
  2005
• Measures clinical and non-clinical achievement
• Focus on long term conditions
• Prevalence proportion of patients on disease
  register
• No demographic data so crude prevalence only
• Data freely available from the Information Centre
  website http//www.ic.nhs.uk

4
Uses of Chronic Disease Registers

• Chronic disease register is basis of recall
  system
• Provides indication of number who need service
• Location of the people who need services
• Available through the QOF
• I will focus on two chronic diseases
• Coronary Heart Disease
• Diabetes

5
Prevalence of Diabetes CHD 2004/05
Data from QOF
6
Is QOF Prevalence Reliable?

• QOF prevalence may not reflect real prevalence
• Registers may be inaccurate or incomplete
• Over counting, miscoding of diagnosis
• Undiagnosed, diagnosis not recorded
• Particularly for diabetes
• Care may be managed in hospital
• Type 2 diabetes can be present for several years
  before diagnosis

7
Importance of Reliability

• Standard of care lower in practices with
  inaccurate registers?
• Service redesign could increase inequity
• Inverse care law
• Need to assess QOF register sizes
• Look for practices whose crude prevalence is high
  / low?
• Does not account for demographic differences

8
Prevalence and Age
Data from PRIMIS
9
Comparing Prevalence

• Adjust for demographics of practice population
• Cannot directly standardise as information not
  available
• Alternative Standardised morbidity Ratio (SmR)
• Three steps
• Estimate expected register size
• SmR (Observed register) / (Expected register)
  x100
• Present as funnel plot showing 95 and 99
  confidence range

10
Estimating Prevalence of CHD

• Reference source
• Key Health Statistics from General Practice 1998
• Doctor diagnosed and treated CHD
• Male and female prevalence
• Ages 0-34, 35-44, 45-54, 55-64, 65-74, 75-84,
  85
• Registered patients by age and sex from PSU
• Multiply age and sex specific prevalence by
  number of patients to get estimated register size

11
CHD Funnel Plot
12
SmR and Deprivation
r2 0.18
13
CHD Comparison to SHA
14
Estimating Prevalence of Diabetes

• Source of prevalence estimate
• PBS model
• Diagnosed AND undiagnosed diabetes
• Input into model
• Registered patients by age and sex from PSU
• Ages 16-24, 25-34, 35-44, 45-54, 55-64, 65-74,
  75
• Black and Asian estimated from 2001 census
• Deprivation adjustment for ward which has the
  highest numbers of resident patients
• Model outputs estimated register size

15
Diabetes Funnel Plot
16
Diabetes Comparison to SHA
17
Conclusions

• SmR gives indication of the completeness and
  accuracy of chronic disease registers
• QOF data indicate that registers record
  approximately
• 84 of the predicted number of people with CHD
• 80 of the predicted number of people with
  diabetes
• Incomplete registers were expected for diabetes
  but not necessarily for CHD
• Limitations
• Accuracy of expected prevalence models
• Incomplete adjustment for population
  characteristics

18
Contact Details

• Jennifer Nicholas
• Public Health Intelligence Epidemiologist
• Hertfordshire PCTs
• Charter House
• Parkway
• Welwyn Garden City
• AL8 6JL
• Phone [phone] ext 2410
• Email jennifer.nicholas_at_herts-pcts.nhs.uk
• Website www.phi-bedsherts.nhs.uk