Alternative population sampling frames produced important differences in estimates of association: a case-control...

Area of Science:

• Epidemiology
• Health Services Research

Background:

• Health service registers are common population sampling frames in universal healthcare systems.
• Commercial databases offer an alternative with claimed large population coverage, ease of access, and flexibility.

Purpose of the Study:

• To evaluate the use of a UK health service register against a commercial database for population controls in a case-control study.
• To compare the characteristics of controls recruited from National Health Service (NHS) primary care practices versus a commercial database.

Main Methods:

• A case-control study of vasculitis was conducted in Scotland.
• Controls were recruited from NHS primary care registers and a commercial database.
• Characteristics of controls and case-control associations were compared using logistic regression.

Main Results:

• Participation rates were 88% for cases and 24% for both control groups.
• NHS controls reported poorer health (physical component score, chronic widespread pain, fatigue) compared to commercial database controls.
• Differences in control health influenced association estimates, with one significant association (pain) observed with commercial controls but not NHS controls.

Conclusions:

• Significant differences exist in self-reported health and quality of life between controls from NHS registers and commercial databases.
• Methodological rigor and prior assessment of sampling frames are crucial for case-control studies.
• The choice of sampling frame can impact study findings and requires careful consideration.