Table 1.

Specification and emulation of a target trial of 3 cycles of IUI versus one cycle of ART and pregnancy outcomes using IBM MarketScan Commercial Claims and Encounters Database, United States, January 2011-Oct 2015.

	Target trial specification	Target trial emulation
Eligibility criteria	• Women aged 18–45 years • With an infertility diagnosis • No history of IUI or ART for at least 12 months • Enrolled in MarketScan for at least 12 months • Insurance coverage for IUI and ART treatments • No pre-existing diabetes or hypertension A variation of target trial would require no history of IUI or ART for at least 18 months.	Same.
Treatment strategies	1. one cycle of ART immediately, no more cycles of ART or IUI for 4 months 2. one cycle of IUI immediately, 2 additional consecutive cycles of IUI within 4 months unless pregnancy occurs, and no ART within 4 months. Cycles are considered consecutive if they occur within 60 days of each other. Women would be excused from an additional IUI if/when they develop serious medical conditions (e.g., cancer, psychosis) within 4 months from first cycle. • Under both strategies, there would be specific instructions to avoid spontaneous pregnancy during cycles without fertility treatment. • Under both strategies, ovulation stimulation protocol performed as clinically indicated; a pregnancy test would be performed 2 weeks after an IUI or embryo transfer and ongoing pregnancy would be confirmed by ultrasonography. • ART includes in vitro fertilization and intracytoplasmic sperm injection without preimplantation genetic testing. Insemination, fresh or frozen embryo transfer, and number of embryos transferred as clinically indicated. A variation of the target trial would have no instruction to avoid spontaneous pregnancy.	Same. We defined baseline as the first procedure date of IUI or ART in the Marketscan. Because pregnancy test results are not systematically recorded in the database, we identified pregnancy by the presence of end of pregnancy outcome code (abortion, termination, stillbirth, or livebirth) in the database.
Treatment assignment	Participants are randomly assigned to either strategy and are aware of the strategy to which they have been assigned.	We assumed that women were randomly assigned within levels of baseline covariates: age, calendar year of the cycle, infertility diagnosis, polycystic ovarian syndrome, overweight or obesity, and region of residence.
Outcomes	• Pregnancy outcomes: pregnancy, live births. • Maternal outcomes: gestational diabetes (GDM), preeclampsia (PE), gestational hypertension (GHTN). • Neonatal outcomes: multiple births, non-chromosomal congenital malformation, preterm birth, small-for-gestational-age (SGA), large-for-gestational age (LGA), NICU admission.	Same. Pregnancy outcomes were ascertained by ICD-9 and CPT codes.
Follow-Up	Follow-up starts at assignment to a strategy and ends at live birth, 14 months after randomization, loss of insurance eligibility, or October 1, 2015, whichever comes first	Same
Causal contrasts	Intention-to-treat effect. Per protocol effect.	Observational analog of intention-to-treat and per protocol effect.
Analysis Plan	Intention-to-treat analysis: For neonatal outcomes, compare the probabilities of live birth overall, and live birth with and without an infant event, under each strategy. For maternal outcomes, compare the probabilities of pregnancy ≥ 20 weeks with and without a maternal event. Per-protocol analysis: Same as the intention-to-treat analysis, except that women are censored (and thus excluded) if/when they deviate from protocol Subgroup analyses by age (18–40 years versus 41–45 years), by insurance plan, in women living in states with insurance mandate, and in women with unexplained infertility.	Same intention-to-treat analysis with additional adjustment for baseline covariates. Due to high levels of non-adherence, intention to-treat analysis was conducted for pregnancy outcomes only. Same per-protocol analysis. Same subgroup analyses. “Unspecified origin of infertility diagnosis” in the ICD-9 code was taken as unexplained infertility.