Realizing the full potential of registry-based randomized controlled trials

Advantages of registry-based randomized controlled trials (RRCTs) include their relatively low cost and ability to recruit many individuals. Notably, the registry infrastructure captures data on all individuals, which facilitates a direct comparison of the baseline characteristics of enrolled and non-enrolled individuals. Decision-makers can then assess how representative the trial population is of the target population to which the results are intended to be applied.

There are two important factors that must be considered. First, a potentially overlooked issue is the differing proportion of missing baseline data for enrolled and non-enrolled individuals. This discrepancy may be attributable to the individuals' underlying clinical condition, their level of healthcare engagement, or their clinicians' awareness of their enrollment status. The data are therefore likely to be missing not at random, and comparisons of baseline characteristics may not be valid. Second, it is important to consider whether non-enrolled individuals would be eligible for the trial's target population, where only clinically ineligible individuals should be excluded. Individuals who are not enrolled because they do not or cannot consent, or were not asked, should therefore be included. However, many trials, including RRCTs, do not always explicitly report the reasons for non-enrollment.

To take full advantage of RRCTs, we recommend (1) reporting the proportion of missing data for baseline characteristics of both enrolled and non-enrolled individuals, (2) reporting the reasons for non-enrollment, and (3) restricting comparisons of baseline characteristics to those enrolled in the trial and those who would be in the trial's target population.

Health-care registries are organized databases that collate information on individuals with specific diseases or conditions. Randomized controlled trials can be conducted within existing registries, which are known as registry-based randomized controlled trials (RRCTs). The registry infrastructure means that information can be captured both on individuals who are and are not enrolled in a trial. This allows an assessment of how representative the trial population is of the group of individuals to whom the results are intended to be applied. However, two important factors must be considered. First, the amount of missing data for the reported baseline characteristics can vary between enrolled and non-enrolled individuals. It is likely that there are underlying reasons driving this discrepancy. For example, an individual may be too unwell to consent to enroll in a trial and too unwell to provide baseline information. The varying proportion of missing data may therefore mean that it is not a fair comparison of baseline characteristics between enrolled and non-enrolled individuals. Second, randomized trials do not always report the reasons why individuals are not enrolled. This is important in understanding if the results of the trial can be extrapolated to the non-enrolled population. The non-enrolled group should include only individuals who did not or could not consent, or who were not asked, but not individuals who cannot receive the treatment for clinical reasons (e.g., due to an allergy). We recommend that RRCTs report the extent of missing data for all individuals, document reasons for non-enrollment, and present information for the non-enrolled group only for those who are clinically able to receive the treatment.