Supplement 1. Results from net benefit regression models for various willingness-to-pay value
* BOOTSTRAPPING
set more off
gen delta_c = tx
gen delta_e = tx

bootstrap, reps(1000) strata(tx) saving("G:\...\bootstrap_c.dta", replace) bca verbose seed(1234) : reg cost delta_c
bootstrap, reps(1000) strata(tx) saving("G:\...\bootstrap_e.dta", replace) bca verbose seed(1234) : reg effect delta_e
clear
use "G:\...\bootstrap_e.dta"
merge using "G:\...\bootstrap_c.dta"
gen delta_c = _b_delta_c
gen delta_e = _b_delta_e
gen icer = delta_c / delta_e
keep icer delta_c delta_e

scatter delta_c delta_e, title("Bootstrapped replicates of ICER estimates") xtitle("Incremental Effect") xline(0) ytitle("Incremental Cost") yline(0) xlabel(-2.5(0.5)1) ylabel(-250(100)250)


* NET BENEFIT FRAMEWORK
rename tot_posttest effect
rename tot_cost cost
replace tx = 0 if tx==2

gen nb0 = 0*effect - cost
gen nb100 = 100*effect - cost
gen nb200 = 200*effect - cost
gen nb300 = 300*effect - cost
gen nb400 = 400*effect - cost
gen nb500 = 500*effect - cost
gen nb1k = 1000*effect - cost
gen nb5k = 5000*effect - cost
gen nb10k = 10000*effect - cost
gen nb20k = 20000*effect - cost

reg nb0 tx training tot_pretest
reg nb100 tx training tot_pretest
reg nb200 tx training tot_pretest
reg nb300 tx training tot_pretest
reg nb400 tx training tot_pretest
reg nb500 tx training tot_pretest
reg nb1k tx training tot_pretest
reg nb5k tx training tot_pretest
reg nb10k tx training tot_pretest
reg nb20k tx training tot_pretest

Note. WTP = willingness-to-pay; PGY = postgraduate training year; ANTS = the Anesthetists’ Non-Technical Skills; TX = self-debriefing; CE = cost-effective (because INB > 0); Not CE = self-debriefing is not cost-effective compare to instructor debriefing (because INB < 0)