# Transporting: What Is It and How Do You Do It?

## TL;DR

This paper explains transportability, a concept in epidemiology that helps apply study results to different populations, and outlines methods to achieve it.

## Contribution

The paper provides a tutorial on transportability methods and distinguishes it from generalizability with formal definitions and analytic approaches.

## Key findings

- Transportability requires defining core conditions for transferring treatment effects to a new population.
- The paper outlines approaches to identify adjustment sets and design estimators for transported effects.
- A case study demonstrates transporting chemotherapy effect estimates from a trial to a cancer registry.

## Abstract

Transportability, one of the twin faces of external validity (alongside generalizability), refers to the ability to use effect estimates in a study population to understand effects in a different population. In this review, we aimed to provide an overview of ongoing methodological developments in the field of transportability and provide a tutorial walking through key steps in the transportability process.

We cover recent work done to distinguish the concept of transportability from generalizability (or external validity more broadly), define core conditions necessary for transporting treatment effects to a different target population, outline approaches to identify sufficient adjustment sets, and design estimators to estimate transported treatment effects. We then illustrate the application of these methods through a case study comparing the effects of two adjuvant chemotherapies for breast cancer in patients within the National Cancer Database, a large national cancer registry, using effect estimates transported from a randomized controlled trial.

While external validity, generalizability, and transportability have long been recognized as important elements of epidemiology, they have historically been treated interchangeably and discussed qualitatively in discussion sections of manuscripts. Over the past two decades, however, major strides have been made to formally define these concepts and introduce analytic methods for them that are valid under well-defined conditions.