Skip to main content
Fig. 3 | BMC Medical Research Methodology

Fig. 3

From: A novel weighting method to remove bias from within-subject exposure dependency in case-crossover studies

Fig. 3

Estimates of OR_SCL, OR_VF, OR_MH and OR_G for simulated data in Fig. 1. OR_SCL, OR_VF and OR_MH with their 95% CI, and OR_G with its 2.5–97.5pct for the exposure (exp(β)) for M = 1, 3, 6, 9, 13, 20, 69, and 139, where M is the number of control periods and study period = M + 1 days (1 period = 1 day). (a) shows results for the population in Fig. 2a with one binary exposure variable only. (b) shows results for the population in Fig. 2b with one binary exposure variable and one binary time-varying confounder. OR_SCL: odds ratio by the standard conditional logistic regression (ORSCL); OR_VF: odds ratio by the Vines and Farrington’s method (ORVF); OR_MH: odds ratio by the Mantel-Haenszel method (ORMH); OR_G: odds ratio by the Greenland’s method (ORG); 95%CI: 95% confidence interval; 2.5–97.5pct: 2.5 to 97.5 percentiles

Back to article page