by Joanne B. Severe, MS, and Annette Stemhagen, DrPH, FISPE
Ms. Severe is a research consultant in Rockville, Maryland, and Dr. Stemhagen is with United Biosource Corp. in Bethesda, Maryland.

Innov Clin Neurosci. 2013;10(5–6 Suppl A):29S–31S

Funding: Funding provided by the ISCTM?for the preparation of this article.

Financial Disclosures: The authors have no conflicts of interest relevant to the content of this article.

Key words: Registries, clinical trials, registry data, augment clinical trials

Abstract: This article is based on the proceedings from a workshop held during the Autumn 2011 International Society for CNS Clinical Trials and Methodology (ISCTM) Conference in Amelia Island, Florida. The goal of the workshop was to establish preliminary steps in determining whether and how patient registries can augment clinical trials in the field of central nervous system therapeutics. Participants in the workshop first defined several different types of registries and then created a list of questions that should be addressed in order to determine how registries might be used. The workshop concluded with discussion on logistical and practical considerations regarding use of patient registries.

Introduction

A workshop to discuss whether and how patient registries can augment clinical trials in the field of central nervous system (CNS) therapeutics was held at the Autumn 2011 International Society for CNS Clinical Trials and Methodology (ISCTM) Conference in Amelia Island, Florida. The workshop was a first step toward the ultimate goal of determining whether registries can provide novel efficiencies in the drug development process or augment, or even substitute for, clinical trial data under appropriate conditions. Participants in the workshop first defined several different types of registries as well as sought to determine what questions would be uniquely suitable to be addressed by registry data that might contribute to the overall database of information for CNS development. The participants also discussed logistical and practical considerations regarding use of patient registries

How is “registry” defined?

The original intent of this workshop was to focus on “large” registries; however, among the group, there seemed to be no single definition of “registry” let alone “large registry.” Thus, participants in the workshop began with a discussion of registries in a broader sense (i.e., What is a registry?) to help focus on the types of registries that would be relevant to the goals. Participants began by generating consensus definitions of the various types of formal and informal observational “databases” that are called registries. Participants defined these registries as follows:

  • Disease registry—A cohort study that follows patients with a specific disease or condition longitudinally to identify risk factors, treatments, disease progression, and mortality. Alzheimers Disease Registry is a well-known example. This type of registry is often used for “orphan” diseases.
  • Disease surveillance, including incidence estimation (e.g., tumor registries)—Identification of new cases to estimate incidence or prevalence; does not follow cases over time other than for mortality. SEER registry is a good example.
  • Exposure registry—A cohort study that follows patients starting a specific treatment longitudinally for outcomes, such as adverse events, quality of life, and healthcare resource use. Usually follows cases over the long term.
  • Risk management program (e.g., clozapine, Sabril)—Enrollment of all patients treated with a specific pharmaceutical/ biologic product into a risk mitigation program to ensure safe use conditions. Additional data regarding adverse events may be collected. Clozapine, Sabril, and Accutane registries are examples of industry-sponsored REMS (Risk Evaluation Mitigation) registries. A pregnancy registry may also fit under this category. These types of registries may follow patients for the development of adverse effects, such as serious opportunistic infection or long-term cancer risks. Such registries often include restrictions on the patient population.
  • Directory of potential clinical trial participants—Identification of patients who may qualify for a clinical trial.
  • Individual databases generated by hospitals based on their catchment area or intake clinics—These usually include case series of patients with specific conditions or characteristics.
  • Data registries attached to biological specimen repositories—Cross sectional or longitudinal data (generally anonymized) that are collected in relation to biological specimen collection. The ADNI registry is an example.
  • Claims databases and electronic health records (not population-based)—Searchable databases used for disease and exposure. Examples may be the Kaiser Healthcare Network, Lifelink Healthcare Database, the VA (Veterans Administration) Database, and the Uniform Data System (UDS) used by the Health Services and Research Administration’s (HRSA) Bureau of Primary Health Care (BPHC).
  • Population-based databases established by countries—Examples are the Israeli Army database and the databases in Sweden.

How might registry data be used?

Clinical trials are usually expensive, take a long time to complete and analyze, and focus on just one (or a few) target questions. Given this, we came up with the following questions during the workshop regarding the role collected data from patient registries might play:

  1. Is there a role for other data sources, when used appropriately, to reduce the time and/or cost of a pre-registration clinical trial, or to provide important post-registration, post-marketing information on safety and effectiveness in real-world settings, akin to the goals of effectiveness trials?
  2. Can data that are collected as part of a registry be used independently to streamline recruitment into a trial (e.g., screen subjects) or to reduce the amount of data collection once enrolled into a clinical trial (e.g., demographics, systematic diagnostic information, medication and treatment course history)?
  3. Can we identify circumstances under which registries can augment or even replace clinical trials?
  4. What characteristics of registries may (or may not) make them useful to clinical trials in CNS?
  5. What are the limitations and strengths of registry data?
  6. Are registries cost effective?
  7. Do registries produce good evidence for effects?
  8. What roles do registry data serve in the pre-approval or post-approval regulatory process?
  9. How can registry data be integrated with other data (observational or data from clinical trials)?

Other Considerations

There are logistical and practical considerations regarding the establishment, maintenance, and access to registry data that were outlined but did not receive discussion in this first meeting. Such issues included management; maintenance; quality control over such things as standards for assessments and standardization of outcome reporting; use of common data elements; legal, ethical, and financial issues; common software; and access to registry data—all of which are critical to evaluating the quality of a registry as well as to utilizing data or individuals enrolled in it. Furthermore, it was noted that using registry patients in a clinical trial then changes the nature of the population of patients in the registry itself.

Conclusion

The workshop concluded that there was no one definition of a registry and that when discussing the merits of this type of observational data, it is important not to use the term “registry data” by itself, but to qualify what kind of registry data is being referenced. A series of topics were laid out that can generally be lumped under the title of “implications for clinical trials.” For example, given that different types of registries obtain and store different types of data (e.g., some are cross-sectional, some systematically longitudinal, and some event-based; some are population-based and some are narrowly defined populations), the challenge is to understand which types of registries can address which types of questions that would intersect with the goals of clinical trials. In other words, how does one design studies using registry data? Would registry data best be utilized for low frequency serious events? What are the appropriate comparators? Statistical considerations are interwoven with the use of registries for purposes perhaps beyond their original intent. For example: What biases are inherent in a registry in terms of who the participants are? How can those biases be made known and thereby evaluated? What are the statistical implications in the sampling? How would the use of registry patients in a clinical trial affect the generalizability of the trial results?

What also evolved from the discussion was an awareness of the variety of reasons why registries are established. Thus, from an industry perspective, companies develop their own registries for drug development or risk management projects pertinent to the goals of the company. From a public health perspective, the wider availability of registry data and the ability to evaluate the utility of those data for contribution to larger scale effectiveness questions might prove enormously valuable. These two aspects, while recognized, were not addressed in this forum.

This brief workshop raised many issues and answered few regarding whether the use of patient registries would be beneficial in CNS drug development. It is clear that the issues raised span the full range of methodology from defining the questions that can be addressed, to collecting and accessing data, to statistical analytic techniques and assessment for bias. Interested ISCTM members may be asked to form a workgroup to continue the discussions with the goal of finding the sweet spot for the intersection of registry data and clinical trials.