TTE methodology: mapping to reference papers

Overview

This vignette maps the swereg target trial emulation (TTE) implementation to five reference papers that define the methodological foundation. It documents which methods are implemented, which are not, and the rationale for design choices.

Reference papers

1. Hernán & Robins (2016) — “Using big data to emulate a target trial when a randomized trial is not available.” Am J Epidemiol. Theoretical framework defining the four emulation failures and their solutions.

2. Hernán et al. (2008) — “Observational analyses of the effect of combined estrogen-progestin therapy on coronary heart disease.” Epidemiology. The original “sequence of nested trials” paper. NHS/WHI hormone therapy example.

3. Danaei et al. (2013) — “Statins and coronary heart disease events.” Epidemiology. Most detailed methods paper covering ITT, per-protocol, and as-treated analyses with SAS code.

4. Caniglia et al. (2023) — “Emulating target trials in pregnancy.” Am J Epidemiol. Discusses enrollment period granularity and residual immortal time bias.

5. Cashin et al. (2025) — “TARGET Statement.” JAMA. 21-item checklist for transparent reporting of target trial emulations.

Method mapping

Enrollment and trial construction

Method	Paper	swereg implementation
Sequence of nested trials	Hernán 2008, Danaei 2013	TTEEnrollment$new(..., ratio=): Band-based enrollment creates sequential trials at period_width-week intervals.
Cloning + censoring	Hernán 2016	Not used. Our per-band matching is an alternative to cloning that is computationally more efficient for large registries.
Grace period	Hernán 2016 (Section 4.4)	Not implemented. A true grace period (initiation allowed within X weeks of assignment without deviation) requires cloning + censoring + weighting. period_width provides only within-band slack for the timing of initiation at enrollment – treatment starting in any week of the entry band counts as baseline initiation (with the residual within-band immortal time discussed by Caniglia 2023) – and deviation censors at the first mismatched band thereafter. See vignette("tte-nomenclature").

Weighting methods

Method	Paper	swereg implementation
Baseline IPW (propensity score)	Hernán 2008, Danaei 2013	$s2_ipw(): Logistic regression P(A=1 | L_baseline), stabilized by default.
Time-varying IPW (as-treated)	Danaei 2013 (Section 4.3)	Not implemented. Requires P(A_t | A_{t-1}, L_t).
IPCW for per-protocol	Danaei 2013 (Section 4.2)	$s4_prepare_for_analysis(estimand = "pp"): Censoring at protocol deviation, IPCW-PP weights via GAM/GLM. Combined weight: analysis_weight_pp = ipw × ipcw_pp.
Baseline IPW for ITT	Hernán & Robins	$s4_prepare_for_analysis(estimand = "itt"): no switch censoring, no IPCW; analysis weight = baseline ipw_trunc.
Weight stabilization	Danaei 2013	IPW: stabilized with marginal treatment probability. IPCW-PP: simplified stabilization using marginal censoring probability (see note below).
Weight truncation	Danaei 2013	$s3_truncate_weights(): Winsorization at 1st/99th percentiles by default.

Note on analysis types: swereg supports both per-protocol and intention-to-treat estimands, selected with $s4_prepare_for_analysis(estimand = "pp" | "itt"). Per-protocol censors follow-up at treatment switching and corrects the resulting informative censoring with IPCW (analysis weight analysis_weight_pp[_trunc]). Intention-to-treat keeps follow-up through switching — no switch censoring, no IPCW — and weights on the baseline IPW alone (ipw_trunc). The production pipeline builds both analysis files per ETT and reports both IRRs side by side. As-treated analysis (time-varying IPW) is not implemented.

Follow-up can end for five reasons; only treatment switching is handled differently between the two estimands:

Reason follow-up ends	Per-protocol	Intention-to-treat
Outcome event	ends, counts as event	same
Outcome event AND switch in the same band	counts as event (event-priority, since 26.7.3)	counts as event
Treatment switch (protocol deviation)	censors, IPCW-corrected	ignored (not censored)
Loss to follow-up (records end early)	censors, IPCW-corrected	censors, treated as independent
Administrative end (study cutoff)	censors	same
Follow-up horizon	censors	same

Per-protocol weight = ipw × ipcw_pp; intention-to-treat weight = baseline ipw only.

Note on IPCW stabilization: Danaei (2013) describes stabilized IPCW weights with a numerator conditioned on baseline covariates. Our implementation uses the simpler marginal (population-average) censoring probability as the numerator. This is a common simplification that is equivalent when baseline covariates have limited predictive power for censoring.

Outcome models

Method	Paper	swereg implementation
Cox proportional hazards	Hernán 2008 (ITT)	Not directly. $km() provides IPW-weighted Kaplan-Meier curves via survey::svykm().
Pooled logistic regression	Danaei 2013, Hernán 2008 (IPW)	$irr(): Weighted Poisson regression with survey::svyglm(family = quasipoisson), computationally equivalent.
Flexible baseline hazard	Danaei 2013 (“month of follow-up and squared terms”)	$irr(): splines::ns(tstop, df=3) models the baseline event rate flexibly.
Trial as covariate	Caniglia 2023, Danaei 2013	$irr(): Includes trial_id in both outcome and IPCW models (ns for ≥5 trials, linear for 2-4).
Robust variance	Hernán 2008, Danaei 2013	survey::svydesign(ids = ~person_id_var) provides person-level clustered standard errors.
Heterogeneity test	Hernán 2008, Danaei 2013	$heterogeneity_test(): Wald test on trial_id × treatment interaction.

IRR approximates HR

For rare events (typical in registry-based TTE studies), the incidence rate ratio from Poisson regression approximates the hazard ratio from Cox regression (Thompson 1977). The quasipoisson family in svyglm additionally accounts for overdispersion from survey weights. This approach scales to large registry datasets where survey::svycoxph() would be computationally prohibitive.

TARGET checklist

The TARGET Statement (Cashin et al., JAMA 2025) provides a 21-item reporting checklist. Use plan$print_target_checklist() to generate a pre-populated checklist that maps the 21 TARGET items to swereg configuration, showing what’s auto-populated from the spec and what needs manual reporting.

What is not implemented

Method	Reason
As-treated analysis (time-varying IPW)	Requires different modeling framework for time-varying treatment weights. May be added in future versions.
Log-binomial models (risk ratios)	Caniglia (2023) uses these for pregnancy outcomes. Users can fit these externally on $extract() data.
Baseline-conditional IPCW stabilization	Requires fitting a second censoring model for the numerator. Marginal stabilization is sufficient for most applications.