Use of Historical Controls in Clinical Trials

INTRODUCTION

This JAMA Guide to Statistics and Methods explains the use of historical controls—persons who had received a specific control treatment in a previous study—when randomizing participants to that control treatment in a subsequent trial may not be practical or ethical.

A randomized clinical trial (RCT) is frequently the preferred research design for testing new medical treatments. Randomization helps to ensure that the participants in the treatment groups are similar in the distribution of prognostic factors.¹ This minimizes bias in statistical comparisons of patient outcomes and allows differences to be interpreted as the causal effect of treatment assigment.² However, there are situations in which randomizing participants to a control treatment or placebo within an RCT may not be practical or ethical, eg, if a comparison with placebo is desired but an effective treatment already exists. In such cases, researchers might use data from participants who had received the intended control treatment in a prior study, termed historical controls, to estimate the benefit of the new treatment.

In an article published in JAMA, Coelho and colleagues³ presented the primary results from the NEURO-TTRansform trial, a clinical trial of eplontersen treatment in patients with hereditary transthyretin amyloidosis. The primary analysis compares the outcomes of participants assigned to receive eplontersen with those from an external control population of participants administered a placebo in a prior clinical trial.⁴ There were no concurrently enrolled patients assigned to placebo in the NEURO-TTRansform trial, necessitating the use of historical controls from a prior clinical trial to enable a comparison of eplontersen with placebo.

USE OF HISTORICAL CONTROLS

What Are Historical Controls?

The term “historical controls” refers to a group of patients drawn from a previously existing population, rather than one generated through concurrent enrollment and randomization within a clinical trial. The outcomes of the current participants treated with the therapy of interest are compared with those of the prior, historical control patients. The historical controls could be patients from a previous clinical trial, an existing clinical registry, or even identified from chart review of medical records.

Why Are Historical Controls Important?

Making a randomized comparison with a placebo treatment in a clinical trial is not feasible in all clinical settings, eg, when studying diseases with well-established, effective treatments when randomization to placebo is unethical.⁵ Historical controls are an alternative to randomization to a placebo in these settings, providing a comparison that is helpful in understanding the clinical effect of an intervention.

Historical controls can also be useful in other settings. In the setting of rare disease, where recruitment is inherently difficult, historical controls may be used in lieu of a randomized comparator to reduce the size of the trial.⁶ Historical controls are also used for diseases with particularly severe outcomes or that ...