Use of Whole Population Registers:

1
Use of Whole Population Registers

• Advantages and Disadvantages

2
Problems in Observational Studies

• Who gets included?
• Who gets lost?
• How to ensure completeness of recruitment?
• How to standardise assessment?
• How to collect routinely gathered data fit for
  purpose?
• Source of control group(s)?

3
Threats

• Internal validity
• External validity

4
Problem withmy clinical series

• Who are your patients?
• Which patients could you have included?
• Which patients have you lost
• And why?

5
Issue of Catchment population
Your area
Your Clinic
6
Issue of Catchment population
Other clinics
Your area
Your Clinic
7
Issue of Catchment population
Your area
Your Clinic
Other Clinic
8
Issue of Catchment population
Your area
Your Clinic
Other Clinic
???
9
Issue of Catchment population
Your area
Your Clinic
???
Other Clinic
10
Does it matter

• May or may not?
• Selection factors related to
• Disease severity
• Access
• Costs
• Education
• Co-morbidity
• Waiting time etc

11
Who gets lostThe issue of left censorship
12
Recruiting patients from clinic
2007
2001
13
Recruiting patients from clinic Attenders
between January 2003 and December 2004
2007
2001
14
Recruiting patients from clinic Attenders
between January 2003 and December 2004
2007
2001
15
Recruiting patients from clinic Attenders
between January 2003 and December 2004
2007
2001
16
Recruiting patients from clinic Attenders
between January 2003 and December 2004
2007
2001
17
Recruiting patients from clinic Attenders
between January 2003 and December 2004
? Died ? Remitted ? Lost hope
2007
2001
18
What is the message?

• Recruiting current attenders is biassed
• towards
• Survivors
• Continuing problems
• Specific socio-economic groups
• Treatment responders/non-responders
• People who like you!

19
Ideal

• Whole population
• Captured at time of onset (inception cohort)

20
How to ascertain cases from whole population?

• Fix population
• Health plan coverage
• Other special group (eg Nurses)
• Geographical
• (beware of selection factors for 1 and 2)

21
Tertiary Care
Secondary Care
Primary Care
Self Care
22
Threshold vary

• Disease severity
• Socio-economic/education
• Availability of care
• Psychological factors

23

• Minimum entry severity point has to be
• primary care
• BUT
• Still legitimate to use other cut offs if
  external validity

24
Choices for ascertainment

• Detect diagnosed cases based on database search
  and chart review
• Administrative database (eg Pharmex, GPRD)
• Institutional database (eg Mayo Clinic)
• Set up prospective system

25
Use of diagnosed cases

• Cheap
• Quick
• Will allow retrospective recruitment
• Not relying on compliance

26
Prospective system

• Accuracy of data
• Reliability of data
• Timeliness of data
• Build in appropriate follow up
• ?consent/ethics

27
AttritionLosses from cohort

• Why
• Die
• Get better
• Deteriorate (DNA)
• Lose interest
• Change doctor
• Move

28
In practice

• Losses from cohort are greater threat to
• validity that failure to recruit

29
Minimise attrition

• Engaging subjects with research
• Frequent contact
• Feedback
• Consent
• Baseline data on key informants
• Consent to access medical and other records
• Linkage to other datasets
• Subjects do opt in