Table 2.

Findings of studies investigating timing of dialysis initiation; target trial emulation finds results similar to the randomized trial, whereas previous analyses were biased in either of two ways

Specific Analysis	Correct Study Design	Biases Introduced	Confounding Adjustment Necessary	Hazard Ratio (95% CI) for Early Versus Late Dialysis Initiation
Randomized IDEAL trial	Yes	—	No	1.04 (0.83 to 1.30)
Target trial emulation analysis	Yes	—	Yes	0.96 (0.94 to 0.99)a
Common but biased analysis 1	No	Selection bias (depletion of susceptibles), lead time bias	Yes	1.58 (1.19 to 1.78)b
Common but biased analysis 2	No	Immortal time bias	Yes	1.46 (1.19 to 1.78)b

Based on the study by Fu et al.⁸ CI, confidence interval.

^aNote that point estimates between IDEAL and target trial emulation analysis are virtually identical (1.04 versus 0.96).

^bHazard ratios obtained with traditional analyses are similar in magnitude to findings from previous observational studies that used the same biased study designs.