How Should We Read Clinical Trials?
Designing, executing, and interpreting randomised clinical trials.
While a large body of details is required to master the mechanics of a clinical trial, it is just as important to ensure researchers consider the “big picture” of clinical research, Professor Timothy Jackson told a recent conference.
In a presentation discussing the design, execution, and interpretation of randomised clinical trials, he stressed the importance of thinking about papers and bringing critical appraisal to clinical research—rather than just absorbing the facts.
Clinical trials are any form of planned experimental study designed to evaluate the effect of a treatment on a clinical outcome. There are two distinct study designs: observational and experimental. Prof Jackson focused on the latter, outlining the process of designing and executing a clinical trial of either a drug or a medical device.
There are four phases of clinical trials. Phase I, often referred to as “first-in-man”, is an experimental study for a new drug or regimen routinely used by “healthy” volunteers, ranging somewhere between 20 and 80 participants. The primary aim is to find a dose with an acceptable level of safety, including the biological and pharmacological effects of a treatment.
For phase II, researchers may recruit an estimated 20 to 300 patients (and controls), including patients diagnosed with a specific condition, to obtain a preliminary efficacy estimate. These trials collect data in each trial arm to potentially design a larger subsequent trial. Phase IIA is routinely used to specify the dose, while phase IIB aims to evaluate efficacy and ongoing safety.
Phase III studies generally comprise large, multicentre randomised clinical trials with a control group ranging from 300 to 3,000 patients. Studies in phase III routinely present a formal hypothesis in advance, the types of statistical tests planned, and the outcomes to collect. The design should confirm preliminary evidence accumulated from previous exploratory studies showing a drug is both effective and has an acceptable safety profile in the intended indication and target patient population.
The fourth phase of a clinical trial transpires after the approval of a drug or device is made available on the market. These studies continue monitoring efficacy and safety on a broader population consistent with the approved labelling and may expand to other groups, such as paediatric patients or perhaps pregnant patients.
Go with PICO
“PICO” is a useful mnemonic tool for formulating efficient questions: Population, Intervention, Comparison/Control, and Outcome. The “population” aims to define the characteristics of the patient groups included in the study—defining basic inclusions and exclusions such as age, gender, and disease severity. If the inclusions and exclusions are too tightly or selectively defined, the subsequent results may not be generalisable for a useful population.
The “intervention” describes the specific treatment studied—a drug, device, or regimen—and considers the treatment delivery method: monthly injection, a slow-release formulation, etc. The treatment or intervention the comparator or control group receives, whether a placebo or an active comparator, or “standard of care”, defines the “comparison”.
Finally, the “outcome” is where the “rubber meets the road”. It is critical to consider how measurements are collected, ensure they are appropriate and clinically relevant, and examine what functional outcomes are useful for the patient. Prof Jackson also noted it essential to define the primary and secondary endpoints, whereby the primary endpoint meets “approval” with at least 90% power and accepted by independent market regulators, and the secondary endpoints have designated use.
He also emphasised PROMs—patient-reported outcome measures. While these are routinely considered secondary endpoints, they are extremely important in the context of novel treatments and comparing other available options.
As an example PICO question, a search could identify an interventional RCT clinical trial (the study design) of an AAV gene therapy (the intervention) for a specific retinitis pigmentosa patient (the population) to improve the disorder (the outcomes), given no other available treatments (the comparison).
A clear methodology should be available when executing a study so an independent evaluator can assess the given protocol and ensure everything follows the plan. They can subsequently show how the study went, and if not to plan, why and how researchers addressed the deviations.
Prof Jackson emphasised the concern of bias in trial design. Many common sources of bias need active monitoring, including selection bias, performance bias, attrition bias, and reporting bias. And in reading any paper, he stressed, ask if bias strengthens or weakens the results in the context of the “bigger picture”.
At the results stage, it is essential to consider what is meant by statistical significance versus clinical significance. For example, if a paper reports a treatment difference for a logMAR of 0.1 visual acuity, compared to a treatment difference for a logMAR of 0.3 visual acuity, then with all other things equal, the latter logMAR 0.3 difference may likely be relevant for a patient.
Prof Jackson spoke at the 22nd EURETINA Congress in Hamburg.
Timothy L Jackson PhD, MB ChB, FRCOphth is a Professor of Retinal Research and Consultant Ophthalmic Surgeon at the School of Life Sciences and Medicine, King’s College London, UK.
Thursday, March 2, 2023