Real-time registries: the future basis of observational studies?

My aim here is to develop the idea of a real-time registry as a natural improvement upon the traditional idea of a disease registry, one that should expedite medical research and improve patients’ lives.

Observational studies involve a variety of study designs, data sources and data collection systems. The one thing that unites them all is that the study does not involve any interventions on patients that would not otherwise take place as part of routine care. 

Observational studies can be retrospective or prospective, use electronic health records or insurance claims data, and involve active data collection systems that parallel those of interventional studies, or passive systems that aggregate routinely collected administrative or clinical data.

One stalwart data collection system is the disease registry, capable of sustaining many registry-based study designs and generating real-world evidence that may not otherwise be possible or cost-effective. 

The EMA defines a disease registry as a registry of patients “whose members are defined by a particular disease or disease-related patient characteristic, regardless of exposure to any medicinal product, other treatment or a particular health service”. I take it that this is largely an unobjectionable definition. 

Increasingly, regulators are making conditional drug approvals. These may require a post-marketing safety study, or a post-marketing efficacy study looking at more patient oriented outcomes than those reaching significance in clinical trials (for example, the recent FDA approval of aducanumab for Alzheimer’s disease). It is in these situations that the disease registry comes into its own.

Apart from such post-marketing studies however, there are numerous other ways disease registries can play an important research role. The following additional examples are provided in draft guidance from the EMA:

  • Epidemiological studies of incidence, prevalence, outcomes, prognostic factors and confounding variables.
  • Assessing the size of a target population for a clinical trial and profiling the population in terms of comorbidities, demographics, etc.
  • Pharmacoepidemiological studies examining drug utilisation and dosing.
  • Health economic studies examining utilisation of health care resources such as number of visits, hospitalisations, or laboratory tests performed.

All of this demonstrates the utility of disease registries and registry-based studies. But current disease registries are cumbersome beasts indeed.

They’re time-consuming and expensive to establish, often requiring the co-ordination of many different hospitals or other facilities, a dedicated IT infrastructure for the secure transmission and collation of data, expertise in data cleansing and curation, and the ongoing operational costs to maintain a system that may have a lifetime of many decades. 

How can researchers get the benefits of a disease registry without the cost and effort in setting one up? The answer is what I call a real-time registry. 

The truth is, most if not all of the patient-level data of interest in a disease registry is already recorded in various hospital databases (and perhaps primary care systems also). Namely, electronic medical records and associated data in departments such as pharmacy and radiology. 

Collectively, let’s call all these the electronic health record (EHR). 

The use of EHRs for observational study is not a new idea, and indeed already form the basis of disease registries. But what if instead of a single pull of data for an ad hoc observational study, or a regular review of records for the creation or updating of registry records, the EHR was itself the patient-level record in the registry. 

This is the core proposal here: to bring together EHRs from multiple treatment centres, to be integrated into a unified view that is then updated in real-time. This unified dataset can then not only have all of the analytic capabilities of a registry, but with real-time analytic modules, it can do better. 

Traditionally, the detection of safety or efficacy signals would be performed at pre-specified intervals or after a certain level of power is attained, or perhaps some other criteria. Whatever it is, the point is, it’s often not done in real-time. 

With a constant flow of EHR data, and an analytical module sitting atop the unified dataset, continuously (or something approximating it) monitoring the incoming data flow, such signals can be detected as soon as the criterion for significance is reached.

Similarly, for other requirements, such as monitoring trends in treatment patterns, case load, presenting characteristics and so forth, other analytical modules can provide up-to-the-minute descriptive statistics and be continuously monitoring for pre-defined thresholds to be reached. 

For the population-based epidemiologist too, if the dataset is representative, important signals indicating a change in underlying disease risk, or of a factor being confirmed as prognostic, can all be generated in real-time with the right analytics.

All of this is possible with real-time registries. We’ve made it possible at Real World Health in various ways. 

First, with a common data interface, we can pull together EHRs from multiple sites using multiple systems, into a single unified view. 

Second, with a platform architecture, we can add or remove analytical modules easily, as the analytic requirements evolve over time. 

Finally, with an established network of hospitals, the process of gaining consent for data acquisition can be streamlined according to each hospital’s specific data acquisition process. 

At Real World Health, we aim to be able to set up a real-time registry, even for rare diseases requiring a large network of contributing treatment centres, within 12 months, and often less. 

We also want to make the resulting data more accessible, by using a platform architecture and adding a dashboard interface, all that’s needed is a software license. 

We hope the real-time registry concept will gain traction among both industry and academia, and thereby expedite medical research, especially into rare and orphan diseases, and have a measurable impact on patients’ lives.