Simplify data collection, utilising information collected during routine clinical care and minimising collection of additional (non-critical) elements

Require utilisation of objective endpoints

Less complex in terms of regulatory documents (eg, ICH-GCP guidelines)

Efficient use of existing infrastructure and personnel resources

Address problem of allocation bias inherent in observational registries

More likely to reflect generalisable population

Less restrictive enrolment criteria increase rate of recruitment

Permit study of clinical questions unlikely to be funded from other sources (eg, evaluation of therapies that are inexpensive to society, such as generic drugs)

Faster implementation of results

Less expensive to conduct than typical randomised trial

Facilitate pretrial feasibility studies

Usually open-label; blinding may reduce efficiencies

Endpoints are limited to those less subject to bias (eg, all-cause mortality); blinded adjudication committees are necessary to reduce potential bias

Less certain about validity of baseline and outcome data

Less practical for trials requiring extensive data collection or involving centres without reliable electronic health records or centralised event-reporting systems; require consistent health informatics systems

Multicentre/multiregional studies can be difficult to conduct due to interoperability of data systems, harmonisation of data dictionaries or other data compatibility issues

Monitoring and validation still required

Concerns about acceptance from funders, reviewers, editors and regulators