Table 1.

Comparison of the estimated causal odds ratio using the trend-in-trend design and the cohort study method. The population is stratified into five subgroups for the trend-in-trend algorithm. Confounders are sampled only once and fixed over time. CPE indicates cumulative probability of exposure, SD standard deviation.

	Number of Unmeasured Confounders	C-statistic of the CPE Model	Trend-in-Trend Odds Ratio		Cohort Study Odds Ratio
			Mean (% bias)	SD*	Mean (% bias)	SD*
2.5	0	0.68	2.47 (−1.2)	0.0165	2.50 (0.0)	0.0092
2.5	2	0.63	2.45 (−2.0)	0.0170	4.75 (90.0)	0.0091
2.5	4	0.51	2.43 (−2.8)	0.0171	4.80 (92.0)	0.0091
2.0	0	0.68	1.97 (−1.5)	0.0147	2.01 (0.5)	0.0087
2.0	2	0.63	1.95 (−2.5)	0.0153	4.22 (111)	0.0081
2.0	4	0.51	1.94 (−3.0)	0.0131	4.25 (113)	0.0078
1.5	0	0.68	1.52 (1.3)	0.0101	1.50 (0.0)	0.0083
1.5	2	0.63	1.49 (−0.7)	0.0106	3.25 (117)	0.0081
1.5	4	0.51	1.48 (−1.3)	0.0108	3.30 (120)	0.0082
1.0	0	0.68	1.02 (2.0)	0.0082	0.99 (−1.0)	0.0079
1.0	2	0.63	1.02 (2.0)	0.0089	2.08 (108)	0.0074
1.0	4	0.51	1.02 (2.0)	0.0089	2.20 (120)	0.0073