01051nas a2200325   4500000000100000000000100001000000100002008004100003653001100044653006400055653004000119653000900159653001800168653002700186653001900213653002400232653002700256653002200283653004000305653003800345653003300383653002700416100001300443700001400456245019000470250001500660300001200675490000700687020003100694       2024                            d10aHumans10a*Practice Patterns, Physicians'/statistics & numerical data10a*Confounding Factors, Epidemiologic10aBias10aLinear Models10aLeast-Squares Analysis10aUnited Kingdom10aComputer Simulation10ainstrumental variables10anoncollapsibility10aphysician's prescribing preferences10atwo-stage prediction substitution10atwo-stage residual inclusion10aunmeasured confounding1 aL. Zhang1 aJ. Lewsey00aComparing the performance of two-stage residual inclusion methods when using physician's prescribing preference as an instrumental variable: unmeasured confounding and noncollapsibility  a2024/04/03  ae2300850 v13  a2042-6305 (Print)2042-6305