270 likes | 680 Views
Two-stage sampling. JF Boivin Version 14 November 2007. S:BOIVIN695Winter 2007Two-stage Sampling.ppt. 1980s-1990s: Progress in use of administrative drug databases. Advantages. Large Population-based Valid prescription data Long-time periods. Disadvantages.
E N D
Two-stage sampling JF Boivin Version 14 November 2007 S:\BOIVIN\695\Winter 2007\Two-stage Sampling.ppt
1980s-1990s: Progress in use of administrative drug databases
Advantages • Large • Population-based • Valid prescription data • Long-time periods
Disadvantages • Missing data on certain outcomes • Temporal sequence not always clear Glucocorticoids cataracts Cataract surgery glucocorticoids • Lack of data on confounders
Previous research • Poor exposure data Dose Duration Self-reports • Small numbers • Short follow-up • Inadequate control of confounding
NSAIDs and breast cancer • Cases: Saskatchewan cancer registry • Controls: Saskatchewan population • Drug exposure: 15 yr of computerized information • Missing: - Over the counter drugs - Other confounding factors: • Menarche • Menopause • Pregnancies • Obesity
Obese cancer no cancer E+ 2 000 10000 OR=0.5 E− 40 100 10 100 2 040 Not obese E+ 200 10 000 OR=0.5 E− 400 10 000 20 000 600 All E+ 2 200 20 000 OR=2.5 E− 440 10 100 32 740 30 100 2 640 Entire population (= truth)
2 200 20 000 440 10 100 30 100 2 640 Obese cancer no cancer E+ E− Not obese not available E+ E− All E+ computerized databases E−
Option #1 Do not conduct research on that topic
Obese women cancer no cancer ? ? E+ E− ? ? Not obese E+ ? ? E− ? ? All women E+ 2 200 20 000 E− 440 10 100 32 740 Option #2 Cohort or case-control study without data on confounder
Advantages • Cheaper • May be scientifically reasonable for certain questions
Option #3 Collect covariate data on a sample of the study subjects • two-stage samples • three-stage samples • partial questionnaire • case series only • etc.
Two-stage sample Sampling approaches: • simple random • balanced • etc.
227 125 23 2 23 125 227 248 2 200 250/ 250/ 20 000 440 250/ 10 000 250/ (I) 32 740 Two-stage balanced design Obese cancer no cancer E+ E− Not obese E+ E− All E+ E−
White JE. A two-stage design for the study of the relationship between a rare exposure and a rare disease. AJE 1982 Cain KC, Breslow NE. Logistic regression analysis and efficient design for two-stage studies. AJE 1988
Consent for interviews Cases : 49% Controls : 39% (Sharpe et al. Saskatchewan study)
Other related sampling designs • three-stage sampling • partial questionnaire • confounder data on cases only
? ? ? ? 2 200 20 000 440 10 100 30 100 2 640 Confounded data on cases only Obese cancer no cancer E+ 2 000 E− 40 medical record review Not obese 200 E+ E− 400 All E+ computerized databases E−