Table 1.

Types of clinical data available for research studies

Characteristics	Observational data		Clinical trial
	RWD-EHR	Prospective longitudinal cohort study or registry
Definition	Data from EHR relating to patient health status and/or the delivery of healthcare routinely collected from a variety of sources	Non-interventional clinical study, prospectively collecting data on a group of patients with a particular disease or symptom	Patients assigned to one or more interventions to evaluate its impact on healthcare outcomes, for example, randomised controlled trial
Patient population	Broad, encompassing medical system or population area	Restricted by study participation	Restricted eligibility criteria, often excluding elderly and people with comorbidities
Data types	High dimensional	High no, limited by research design and variables for collection decided a priori	Variables/outcomes for collection decided a priori
	Data collected as part of patient care from both patients and physicians	Structured data collection and questionnaires
Data presence	Sparce, noisy	Structured, same data collected on all participants	Highly structured and often with detailed clinical data
	Missingness not at random	Fairly complete	Low missingness
Scale	Large, thousands to millions	Modest, hundreds to thousands	Small, tens to thousands
Generalisability	Strong local structure can restrict generalisability Incorporating real-life noise into the analyses improves applicability to real life settings	Easily replicable in similar designed cohorts Generalisability restricted by patient selection and data not always directly implementable to real life settings.	The more restrained the patient selection the less generalisable

EHR, electronic health record; RWD, real-world data.