2.1. Rare Disease Registry Objectives and Scope

Rare disease registries are initiated by many organizations, such as patients and their families, patient advocacy groups, clinicians, national health systems, and biopharmaceutical product manufacturers, for many reasons. Often, rare disease patient registries have grown organically. In rare diseases where patients are few, research agendas do not exist, standard case guidelines are absent, and patient communities have not yet formed, patient registries are an intuitive first step for stakeholders trying to understand the number of people affected their geographical distribution, and the basic demographic and clinical characteristics of the disease. The scope of these registries may evolve over time, maturing from an outreach/community-building effort or a means for a basic understanding of patient and disease characteristics, to a supportive mechanism for research funding and attracting health care providers. As with all registries, a single rare disease registry need not fulfill all goals for all potential stakeholders. Ideally, however, a well-designed registry provides an infrastructure that can support different needs in an efficient way and eliminate barriers to scientific progress.

It should be noted that rare disease registries include not only diseases that are inherently rare, but also common diseases that are rare in specific populations, especially those defined by demographics. Thus, plaque psoriasis—common among adults—is rare in children, and breast cancer—common among women—is rare among men. While some of the objectives specific to rare disease registries will apply (e.g., patient identification and recruitment), others may not (e.g., disease classification, measuring disease-specific outcomes).

Registries can be developed to serve multiple purposes. The design of the registry depends upon the maturity of the research plan around the disease, the availability and duration of funding, and to some extent, the number of patients affected. For rare diseases, the perception of relative importance of research often correlates with the number of patients affected or the number of empowered disease advocates.

The specific objectives of rare disease registries typically cluster into the following categories:

1. To connect affected patients, families, and clinicians. Patients and families of affected individuals are often interested in knowing about others who share their disease. Many rare diseases have a genetic basis. However, even if multiple family members are affected with the condition, the motivation to be connected to others may be quite strong, driven by their personal desire to know more about the condition, its natural history, alternative coping mechanisms and treatment options, and the diversity of clinical courses and outcomes. The need to connect is enhanced if the patient or family has difficulty in finding an expert to provide advice or the doctor or genetic counselor points out how little is known about the rare condition.
   Registry meetings provide an opportunity to talk and to share personal experiences. These meetings may include lectures and discussion among patients and families and with experts in medicine, dentistry, nursing, sociology, and many other fields. The advent of social media has increased patient involvement in these types of activities by encouraging patient-to-patient dialogue and assisting with recruitment for research and support. Patients and families often want to connect to advocate support of patients' services and financial support for patient care and research.
   Similarly, physicians and other clinicians may want to connect with other clinicians to learn more about the disease and possible treatment options. Most clinicians have not seen a wide spectrum of rare diseases, and little information on some diseases may be available in the literature. Registries may offer a connection to essential information and to experts in the disease area to assist health care providers with advising and counseling patients.
2. To learn the natural history, evolution, risk, and outcomes of specific diseases. Stakeholders often initiate registries to learn the natural history of a rare disease. Typically rare diseases are described in a general way based on their symptoms at the time of diagnosis. With refinement in diagnostic techniques, including genetic, biochemical, and physiological testing, classical disease descriptions are broadened and diseases are better described in terms of the range and likelihood of specific outcomes. Unlike more common diseases, for which criteria for classification will often evolve, such evolution may not be possible for rare diseases, but the acceptance of some general criteria that derive from these studies will inform and help subsequent research. As general and specific therapies emerge, the natural history often changes and the “classical description” may no longer apply. With better therapies for treatment and supportive care, new complications may also be recognized. For treatments that extend life expectancy, what is known about the trajectory of disease can change drastically. A disease registry incorporating patients with rare diseases from many centers allows for gathering stronger and more generalizable safety, diagnostic, and prognostic information.
   For industry, natural history of disease registries are often developed to better understand the burden of disease, elements of disease progression, disease genotypic and phenotypic heterogeneity, and potential endpoints (or surrogate endpoints) that may be used in therapeutic clinical development. Increasingly, these types of disease registries are also used to understand patient and caregiver quality of life and the economic consequences of these diseases, as well as to understand the background risk of specific outcomes (i.e., to provide a reference population) that usually cannot be found or inferred from other sources of data when the disease is rare.
3. To support research on genetic, molecular, and physiological basis of rare diseases. Research on features of disease, both clinical and basic, is a common objective of a registry. Clinical research depends on having a representative population for determining the timing and frequency of natural events and complications, such as development of autoimmune complications, unusual infections, and related or unrelated malignancies. For this reason, rare disease registries benefit from a comprehensive database that is sufficient to address critical clinical questions, while at the same time not being so all-inclusive that the data cannot be acquired and maintained with reasonable effort on the part of the registry team.
   Patients, researchers, and clinicians share interests in understanding diseases at the genetic, molecular, and cellular level. Such studies usually require a biorepository of materials for research, including tissue (fresh and frozen), DNA, RNA, cellular proteins, and bodily fluids. Creating a meaningful repository for the study of rare diseases requires collection of materials from a sufficient population to permit generalization about the fundamental features and diversity of the disease at the genetic, molecular, and cellular level. A registry is an important complement to any biorepository; similarly, biorepositories are far stronger if they are closely linked to a registry that contains relevant longitudinal clinical or phenotype data. In cases where multiple small or regional registries exist for a specific condition, a centralized biorepository can serve as a common link and research resource. Valid interpretation of biosample research depends on understanding the clinical features of the patients and the heterogeneity of the disease in the study population. In addition, the existence of parallel relevant longitudinal clinical data allows for assessment of genetic and environmental disease modifiers.
4. To establish a patient base for evaluating drugs, medical devices, and orphan products. Stakeholders are vitally interested in developing drugs, devices and other therapies for rare diseases, and many rare disease registries have been developed to support the drug development process. Patient registries for rare diseases may emerge from suggestion, pressure, or advocacy of affected patients and/or families. Direct influence can be seen when patients and their caregivers decide they want a registry, raise the funds, and push for its creation. Indirect influence can be seen when patients or special interest groups drive government to make research on that disease a priority. Researchers and industry recognize that a population of patients is essential for clinical testing, and industry may provide rare disease groups with support to begin or expand a fledgling registry so that ultimately a potentially useful drug or device can be tested in the disease population.
   Often, developing a treatment for a rare disease will provide information about pathophysiology that informs treatment development of a related disease. If the rare disease is serious with few or no treatment options, regulators may relax some of the requirements for drug registration (as is indicated by the requirements for orphan drug development).

2.2. Rare Disease Registry Stakeholders

Any registry endeavor has a number of stakeholders, often with both convergent and divergent agendas. Stakeholders may include patient advocacy groups (often multiple), regulatory agencies (especially if the registry is being developed to support future drug development and approval or to fulfill postmarketing commitments or requirements), clinicians, scientists, industry, payers, and the individuals and families affected by the disease. Collaboration between stakeholder groups has been critical to the progress made in research and product development, the adoption of important public policy changes in the United States and worldwide, and the promotion of patient access to treatments as they become available.³ Table 20–1 describes potential registry stakeholders and the roles they may play in registries.

Table 20–1

Role of stakeholders in rare disease registries.

The importance of patient registries in rare diseases and the need to support many organizations has also brought umbrella patient organizations (e.g., NORD, the Genetic Alliance, EURORDIS) in as stakeholders, as these groups are charged with advising and supporting the development of registries. As the number of registries increases along with the number of commercial companies to develop and host them, these umbrella organizations are becoming brokers for services and are motivated to identify standards and shared efficiencies to support patient registries for the thousands of rare disorders that need them. In addition, the proliferation of patient registries for rare diseases brings standards development organizations and standards interests into the fold as the need for standards that can facilitate data sharing (i.e., common data elements) between patient registries and other aspects of health care and clinical research has become evident. More broadly, the vision of patient registries that can share data between electronic health records and personal health records, as well as with clinical research or national public health efforts, has engaged a variety of commercial application providers in the field.

Representatives from any of the groups mentioned as stakeholders can function as registry sponsors or developers. A distinction is made between registry sponsors, as the entities who fund plan, and often select data collection content for a registry, and developers, as the technology and computing professionals who build the registry.

Although data from registries are not a substitute for controlled trials, rare disease registry data may be the only source of information (especially about a specific product's use) available to stakeholders. This information may serve to inform industry such that a controlled trial can be determined to be feasible, designed appropriately, and well informed upon inception. Disease registry data complement trials, especially those conducted in rare diseases, for which other sources of data are rare or nonexistent. Industry supports many rare disease patient registries, both disease-based and product-based, as sponsors and developers. This is particularly common in rare disorders for which the clinical development program is often abbreviated and inclusive of only a small, relatively heterogeneous subpopulation of the disease population. These registries are often well received by patient groups who do not have funds to operate a registry independently, but stakeholder objectives are not always aligned. For example, industry-sponsored registries are in some cases treatment- or product-based registries, where patients are included for study based upon treatment exposure. However, some product exposure registries create a fragmented system that does not allow researchers or policymakers to see the entire spectrum of disease.⁴ These different product registries have different sponsors and collect different data (often at the behest of regulators who seek answers to different questions), rendering them difficult to combine during research. When more than one treatment exists for a given condition, the different postmarketing treatment registries are often not comparable, nor are the full spectrum data (from multiple registries hosted by multiple companies) easily accessible for academic researchers. Additionally, if patients are exposed to multiple treatments, their data might be in multiple registries, but their full experience across treatments is not appreciated.

Disease registries (rather than exposure or treatment registries) create the possibility of assessing the long-term safety and benefit of different treatments, perhaps leading to treatment algorithms that allow more choices for patients and clinicians. Regulators have increasingly recognized the value of disease registries for historical comparator data and long-term evaluation (especially for drug safety) and as a complement to randomized clinical trials to “fill in the blanks” about outcomes that were not addressed in the limited controlled studies. These registries become even more important to regulators (and others involved) when the disease is rare and registries may be the only means by which data can be obtained. The marriage of stakeholder interests may create conflicts of interest for these registries that require careful scrutiny of available resources. If an effective partnership can be established and maintained the creation of clinician and patient/caregiver communities can be a powerful agent in the success of a product in development or evaluation.

Even more effective in rare disease research is a collaborative approach in which multinational and multi-institutional stakeholders combine resources. As resources are combined standardization becomes more important to allow data to be compared across registries. Regulatory organizations such as the U.S. Food and Drug Administration and the European Medicines Agency can guide standardization across multiple postmarket registries within specific disease areas and promote the creation of multisponsor registries where appropriate. Other organizations like PARENT (PAtient REgistries iNiTiative) are facilitating cross-border collaborations to develop comparable and transferrable patient registries. Resources and tools for identifying and sharing patient registry questions, such as the PRISM (Patient Registry Item Specifications and Metadata for Rare Diseases) library of patient registry questions,⁵ will support these standardization efforts.

Although the creation of a single global registry for each disease (or group of diseases) is theoretically a sound idea, in practice it may not always be feasible or in the best interest of researchers. A viable alternative can be a network of registries and resources, such as TREAT-NMD (Translational Research in Europe—Assessment & Treatment of Neuromuscular Diseases), a network of neuromuscular disease researchers that was launched in early 2007. TREAT-NMD aims to create an infrastructure to promote the development of tools (e.g., core outcome sets) that industry, clinicians, and scientists need to bring novel therapeutic approaches through preclinical development and into the clinic and to establish best-practice care for neuromuscular patients worldwide.⁶ Similarly, multiple registries could be connected via a centralized biorepository or biobank to provide larger sample sizes to understand disease processes and how they correlate with patient outcomes. As with any collaborative research, the challenges lie in who manages the collaboration, who funds it, and what governance infrastructure is required to bring together researchers who may be reluctant to share their data. The availability of indices of registries [e.g., OrphaNet⁷ in Europe and the new Registry of Patient Registries (RoPR) in the United States⁸] is helpful for identification of potential data sources and collaborators.

3.1. Patient Population

Because patient registries can collect clinical information from larger, more heterogeneous populations than those included in a clinical trial, they are becoming increasingly valuable, particularly for diseases affecting very small patient populations, such as lysosomal storage disorders and for specific populations such as children.⁹ Whereas selection of patients may be highly restrictive in general disease registries, rare disease registries often have more liberal criteria for inclusion. In many cases, a physician diagnosis, rather than the more common strict classification schema, may be sufficient for inclusion in a rare disease registry. Reasons may include: (1) no classification criteria exist; (2) knowledge of the rare disease is so limited that being more inclusive is desirable; and (3) the population is so small that being more inclusive is desirable. With some exceptions, rare disease registries typically do have broad inclusion criteria and attempt to enroll most, if not all, eligible patients within a targeted geographical area.

Although they may not be sufficient for population-based estimates of disease, these data sources can be used to estimate the numbers of affected patients and the number of patients potentially available for research, and can enable the mobilization of disease-specific communities for advocacy. Since a large proportion of recognized rare diseases are genetic in origin, enrolling family members greatly improves understanding of the disease, but may create additional complexities around confidentiality, logistical issues (e.g., different last names and other tracking issues), and considerations for enrollment of minors. The issue of “study fatigue” should also be considered when developing patient enrollment plans. Because of the limited numbers of available patients, some patients may be asked to participate in multiple studies over time. Patients may become overtaxed by frequent participation in studies and reluctant to join new studies.

For registries examining treatment-related outcomes, the challenges in creating an inclusive patient cohort include differences in health care delivery systems, local regulations, and budgetary considerations that create barriers to care and/or specific treatments. For example, if a disease is rare, a drug or device manufacturer may choose not to go through the rigorous process required to have the product approved or priced locally—for example, in a small country—as the number of patients who might ultimately use the product does not support the cost of time and effort. This may create difficulties in enrolling a representative patient cohort from such regions.

3.2. Data Collection

Most registries are tempted to (and often do) include as many data elements as possible in order to glean as much information as possible from their study population. This often leads to increasing respondent and investigator burden, high rates of discontinuation, and substantial challenges in data management. In some registries, these drawbacks may be offset by the ability to continue to recruit additional patients and/or the availability of sufficient numbers of patients already enrolled despite dropout. This is rarely the case, however, with rare disease registries. Thus, balancing the need for a broad dataset with the burden of data collection is highly important for rare disease registries.

In many respects, data collection for rare disease registries is similar to data collection for other types of registries. Like other registries, rare disease registries aim to collect a uniform set of data on each patient. Data elements should be clearly defined to ensure consistency in interpretation across participating sites, and data collection and management procedures should be designed to support the collection of high quality data. Other chapters in this document discuss these concepts in more detail as they apply to registries generally. However, while many of the best practices described elsewhere in this document are applicable to rare disease registries, rare disease registries face unique data collection challenges not addressed by those best practices. In particular, rare disease registries may encounter additional hurdles when attempting to use common data elements, selecting quality of life or patient-reported outcome measures, collecting biomarkers, obtaining long-term followup data, and assuring data quality.

3.3. Creating Efficiencies in Registry Development

A major step in the development of any registry is the selection of the data elements. This task can be time consuming and resource intensive, particularly when multiple stakeholders are involved in defining the data set. As noted in Chapter 4, the primary goals in selecting data elements are to ensure that the necessary data are collected to achieve the objectives of the study and that the data set is not so overly burdensome as to limit participation in the registry. A critical component of developing the data set is defining the data elements and determining how each piece of data will be collected. Many registries develop and define their own data elements. This approach can be costly, and it limits the ability of data from the registry to be linked or compared with data from other registries or data sources. It is more challenging to standardize the data collection for rare diseases, as the understanding of the disease is likely to be limited and, until recently, established standardization efforts were limited.

Common data elements (CDEs) may offer a potential solution to some of these issues. A CDE can be defined as “a data element that can be consistently collected across all clinical studies.”¹⁰ CDEs include standard definitions, code lists, and instructions so that the data are collected and stored in the same manner by each participating site, in each study. CDEs may be general, meaning they can be used across disease or therapeutic areas (e.g., demographics, vital signs)¹¹ or disease-specific, meaning they are designed for research in a particular disease area (e.g., congenital muscular dystrophy).¹² By using CDEs, registries may be able to reduce the time and effort involved in developing a dataset and to enable the registry data to be linked or compared with data from other studies using the same CDEs.

CDEs are particularly important for rare disease research. CDEs may lower the cost of developing a new registry, making registries more accessible for diseases where funding is limited. CDEs may also enable data from multiple small registry projects to be linked or compared to increase knowledge about the disease. The Institute of Medicine noted the potential importance of CDEs for rare disease research. In the 2010 report, Rare Diseases and Orphan Products: Accelerating Research and Development, the IOM stated “The NIH [National Institutes of Health] should support a collaborative public-private partnership to develop and manage a freely available platform for creating or restructuring patient registries and biorepositories for rare diseases and for sharing de-identified data. The platform should include mechanisms to create standards for data collection, specimen storage, and informed consent by patients or research participants.”¹ Recognizing the potential value of CDEs, NIH recently funded the PRISM project. The objective of the PRISM project, which is administered through the National Library of Medicine and supported by the Office of Rare Disease Research (ORDR), is to develop a library of standardized questions that will be relevant to a broad mix of rare diseases and that can be used to develop new registries or to update existing registries. Ultimately, the project aims to develop tools that will support the rapid implementation of new rare disease registries, the revision of existing registries, and interoperability between rare disease registries and other data sources.^{5, 13}

In January 2010, NIH and ORDR hosted a workshop titled “Advancing Rare Disease Research: the Intersection of Patient Registries, Biospecimen Repositories, and Clinical Data,”^{14, 15} which launched the development of the Global Rare Disease Registry and Data Repository, a Web-based data registry that will link existing registries, future registries, and biorepositories.¹⁶ Two significant work results are expected out of this initiative. The first is a 2-year pilot program in collaboration with Patient Crossroads, Children's Hospital of Philadelphia, and WebMD that will develop a Web-based template to allow any patient group to establish its own patient registry. The second work result is a preliminary set of recommended CDEs¹⁷ that has been drafted and has undergone revision based on feedback received from public comment. The CDEs are generally applicable to any rare disease registry. This CDE list mixes required and optional elements in the following categories: current contact information; sociodemographic information; diagnosis; family history; birth and reproductive history; anthropometric information; patient-reported outcome; medications, devices, and health services; clinical research participation and biospecimen donation; communication and preferences. ORDR is working closely with the Clinical Data Interchange Standards Consortium (CDISC), which has reviewed the ORDR CDEs¹⁸ and is in the process of a similar initiative focused on CDEs for clinical trials. Planned next steps include working with the rare disease community (including clinicians, patients, and advocacy groups) to develop CDEs for specific rare diseases.

While CDEs have significant potential for rare disease registries, they do have some limitations. First, while general CDEs may be relevant for rare disease registries, these CDEs can typically only cover a small portion of a data set necessary for studying a rare disease. The currently available disease-specific CDEs tend to focus on prevalent diseases, such as cancer, cardiovascular disease, and neurological disorders. (Refer to Chapter 4 for a table of currently available CDEs.) Some of these CDEs may be relevant for some rare disease registries, but many may not be useful. Second, CDEs may change over time to reflect changes in practice or new trends in clinical research. Registries with shorter durations (1 to 2 years) may not be affected by changes in CDEs, but rare disease registries are often designed to follow patients for long periods (e.g., several years or until death). To retain the benefits of linkage and comparison, registries that use CDEs would need to update their data collection tools to reflect the changes in CDEs.

In selecting data elements for a rare disease registry, it is useful to consider using CDEs as a first step, before developing new data elements independently. Available CDEs may be relevant for basic information, such as demographics, and for some disease-specific information, and the use of these CDEs is encouraged when possible. When CDEs are not available, a review of the literature and searches of ClinicalTrials.gov and other similar databases may identify other registries or clinical studies in the disease area. Those studies may be willing to share information on their data sets, so that the new registry can either align with those data elements to support future linkages or comparisons or perhaps learn from issues that have come up in the other studies and apply that knowledge to the development of a new data set. As noted in Chapter 4, other considerations in selecting data elements include feasibility (Are the data elements routinely collected in clinical practice?), burden (What scope of data collection is desirable; what burden on participants is tolerable?), and geographic variations in terminology and practice (Do disease definitions differ? Are data collected the same way in all registry locations? Do terminologies vary by country?).

3.4. Including Quality of Life or Patient-Reported Outcome Measures

Quality of life measures and patient-reported outcome measures (PROs) are increasingly being used in registries to understand patient experiences and preferences. In rare disease research, quality of life data and PROs may be particularly important when well-defined widely accepted clinical outcomes are not available. The progression and mechanisms of rare diseases are often not well understood a situation that inhibits the identification of meaningful clinical outcomes, the development of new therapies, and the assessment of the effectiveness of existing therapies or disease management strategies. Quality of life measures and PROs may provide useful data to show that disease management strategies or treatments are effective at improving patient (and caregiver) outcomes or quality of life.

The selection of quality of life measures or PROs for a rare disease registry can be challenging. Disease-specific measures are often not available, and generic measures that were developed with consideration to more common diseases (e.g., the SF-36) are not detailed enough to capture relevant changes in the patient's (or caregiver's) quality of life. New measures may be difficult and expensive to develop, given the small number of patients, validation requirements, and the need to have measures that can be used in multiple languages. As with the selection of data elements, registries may seek to identify other existing studies in the disease area and use similar measures to allow for future comparisons. When selecting measures, burden on the participant is a major consideration. The inclusion of multiple quality of life measures and PROs can be tempting, but they may deter patient participation if the burden is excessive. Considerations for selecting measures, collecting the data, and analyzing and interpreting the data are further discussed in Chapter 5.

In addition to utilizing a PRO tool, a registry may be used to validate one or more PRO instruments across a large number of centers and in some cases countries. Nonvalidated tools, such as patient diaries and other electronic or paper-based data collection methods (e.g., treatment logs to track compliance, symptom checklists), may also be integrated across the registry or within a subpopulation of interest.

Health economic data may also play an important role in a rare disease registry. While the major goals of the registry may be to improve understanding of the disease or to monitor treatments, an underlying objective may be to build the case for new research in the disease area and the development of new therapies. Health economic data may be useful for more broadly demonstrating the global burden of disease. Because of the rarity of the disease, the burden may not be well understood and global burden of disease data may be used to gain support for funding new research in the disease area. In addition, if therapies are developed for rare diseases that were once only treated with supportive care, some aspects of disease burden may improve, while other considerations, such as long-term disease management, may be introduced.

3.5. Biomarkers

Biomarkers, which may describe risk, exposures, intermediate effects of treatment, and biologic mechanisms, are an important component of rare disease research and may serve as important surrogate endpoints for health outcomes.¹⁹ In their report on accelerating rare disease research, the IOM identified biomarkers as an important avenue. When biomarkers have been identified for a rare disease, registries in that disease area should consider collecting biomarker data as part of the registry. Registries in disease areas for which biomarkers have not been identified may also consider collecting biological specimens, physiological tests, or radiographic studies, in the hope of furthering efforts to develop and validate biomarkers.¹

When determining whether to collect biomarkers or other biological specimens, registries must consider several factors, addressing the following questions:

• Does the biosample need to be collected once or on an ongoing basis? If ongoing, how often?
• Does the biosample need to be examined by a central laboratory? Will multiple laboratories be needed because of geographic constraints (e.g., the fact that samples collected in the European Union cannot be sent to a laboratory in the United States) or time constraints (e.g., that sample processing is required within 24 hours)?
• What privacy and ethical issues will the collection and storage of biosamples introduce?
• How will the informed consent document discuss the collection and storage of biosamples? How broad can the language be? Can it address unanticipated use of these samples?
• How long will the biosample data be stored? For what research purposes may it be used?

As more is learned about a rare disease and its origins, the ability to perform exploratory analyses on existing samples is critical and should be considered when a biorepository is established.

3.6. Collection of Followup Data

Collection of long-term followup data is often an important component of debilitating but not fatal rare disease registries. Many rare diseases are chronic and lifelong, meaning that registries may wish to track patients for several years or even until the patient's death. The collection of long-term followup data for rare diseases raises some unique challenges, including what type of providers should participate (specialist vs. general clinician), how to encourage retention, and minimizing lost-to-followup patients.

Many patients with rare diseases see a specialist in the disease area on a regular but infrequent basis (e.g., annually) and see other clinicians on a more regular basis. The specialist may see several patients with the same disease and may have specialized knowledge of the disease; in that sense, the specialist may be an ideal candidate for registry participation. However, participation by the specialist may result in infrequent data collection on the individual patients or missing data collected by other clinicians. The registry may miss events that occur between specialist visits and may not obtain an accurate picture of the day-to-day impact of the disease. Participation by the clinician (or clinicians) that treats the patient on a regular basis is another option. These clinicians may only see one or two patients with the disease and they may not have specialized knowledge of the condition; however, they may be able to provide more frequent updates on the patient's condition. Both of these approaches have strengths and limitations. The most comprehensive approach is to collect data from both the clinician and the specialist. This allows the registry to gather both the specialist's overall perspective on the patient's condition and the more granular details of the patient's care. However, this approach raises privacy issues, as the registry may need to share the data collected from the specialist with the clinician and vice versa. It may also present recruitment issues, as both providers must participate in order to avoid significant missing data. The registry will also need to plan for both physicians to participate in order to avoid having duplicate patients entered into the registry.

Because the collection of long-term data is often critical to the registry's objectives, the registry must devote sufficient effort to patient and physician retention. Over time, patients and physicians may lose interest in the registry and stop participating. Patients who enrolled in the registry as minors may change physicians and locations on multiple occasions and upon reaching adulthood. Patients may move and begin seeing a new physician, or physicians may retire and stop participating. Direct input and access to registry data by affected individuals is technically possible and would allow for ongoing registry involvement and foster retention of patients. It has been observed that patient/family-entered data is reliable information, but that it may not be as in-depth or conveyed in medical language as is the information provided by a physician. Depending on the objectives of the registry, one approach would be to facilitate patient/family entry of data to the patient registry with better design of data entry forms/screens, appropriate contact and followup with participants, and sharing of study results and summary data from the patient registry with participants.

Chapter 10, Section 3 describes many factors that can encourage retention, as well as some potential pitfalls that may impede it. Rare disease registries with long-term followup components should have plans in place to monitor retention rates and should have consistent procedures for when to consider patients lost to followup. Registries that collect sufficient identifiers may also consider linking to other data sources, such as the National Death Index in the United States, to determine if patients have died. Procedures used to locate missing registry participants should be articulated in policies and procedures documents and communicated to participants. In addition, processes to retain patients as their status changes (e.g., from treating clinician to clinician; from minor to adult) should be clearly stated, and multiple contact points should be available for both the individual and their next of kin/designated secondary contact.

3.7. Data Analysis

Patient registries are critical for accruing a sufficient sample size for epidemiological and/or clinical research for rare diseases. In most cases, the registries are not statistically powered for hypothesis testing, and the analytic goals should focus on the descriptive techniques relevant to observational research. The uptake of common data elements could facilitate the formation of analysis data sets from the combination of multiple data sources in situations in which two or more disease registries exist, thus increasing sample sizes.

Data collected in rare disease registries prior to the introduction of therapies that drastically alter standard of care and/or treatment guidelines can provide useful information regarding the natural history of disease. If these data do not exist, removing the effects of widespread treatment(s) from registry analysis on disease progression, particularly in highly heterogeneous disease, is challenging. An additional challenge in rare disease registries is the fragmentation of the data. Patients may contribute data sporadically, but not be completely lost to followup. If data are combined across registries (or other databases), care should be taken to identify potential duplicate patients prior to analysis, as this is more likely to occur in a limited population. Many of the other considerations for analysis (e.g., controlling for confounding, handling of missing data, loss to followup) are not unique to rare disease registry data and are addressed in other chapters, such as Chapters 2, 3, and 13.

3.8. Data Access and Communication

Because populations with rare diseases are often considered vulnerable and under-resourced and because such populations are smaller, it is important to plan for registry data access and communications. Data ownership, data access, and communication are important issues for all registries, but rare disease registries often must pay special attention to these issues because of the broad range of stakeholders involved and the potential interest to others in the disease. Ownership of the data should be clearly specified during the planning phase of the registry and should be communicated to stakeholders and participants in the registry. In many cases, the data owner is the sponsor of the registry. Some registries, though, may have multiple sponsors, or the sponsor may designate that another group will own the data. In addition to ownership of the actual data, ownership of the intellectual property resulting from the registry (e.g., case report forms, patient-reported outcomes tools, reports, analyses, and associated biosamples) should be clearly specified.

Rare disease registries should also develop and adhere to a data access plan. Many data access scenarios are possible. For example, the registry data may only be accessible to the data owner or the sponsor. Alternately, the registry may develop data sharing policies that allow other researchers to access the data. The registry may limit data sharing to investigators participating in the registry, or may allow outside investigators to access the data. For example, an outside researcher may use data from the registry to assess incidence of a particular complication for the purposes of informing protocol development for a new study. As discussed in Chapter 2, data sharing policies should address who can access the data, for what purposes, and under what circumstances (timeframes, access fees, etc.). For example, will manufacturers be able to access data to inform the design of new clinical trials? Will researchers be able to link the data to other data sources for new studies? These types of questions should be carefully considered and addressed in data sharing policies so that all participants (including patients) are aware of the policies and plans for the registry data. Written registry policies and procedures are encouraged and required by many regulatory entities (e.g., institutional review boards).

Publication rights and plans for disseminating information from the registry should also be considered during the planning phase. This is particularly important for rare disease registries with multiple stakeholders, who may have diverse and conflicting interests. There may also be considerations about academic and other interested parties not included in the registry who may wish to use these data at some later point in time. For example, a registry with strong patient advocacy group support and industry funding may need to balance the desire to publish early (in order to share information with the patient community) with the desire to publish later (in order to protect proprietary information related to treatment development). Clear publication plans that are shared with registry stakeholders can help to avoid disagreements once the registry has begun collecting data, and can promote registry transparency.

Like all registries, transparency is important for rare disease registries, and perhaps especially important.¹ Because rare diseases have a limited pool of patients, they need to maintain a highly motivated patient community engaged and actively participating in the registry. Transparency in registry operations, analyses, and publications can help to reassure participants that the registry is fulfilling its objectives and continues to be a worthwhile endeavor. This may take the form of regular updates on registry enrollment and data collection, newsletter updates from principal investigators, and information sharing about abstracts or publications based on registry data. Transparency also requires full disclosure to participants about the use of their data, the registry funding sources, and any underlying goals or motivations for the registry.

3.9. Governance

The governance of a rare disease registry can be extraordinarily simple or, more often, fairly complicated. Complexity stems from the variety of stakeholders involved and their different agendas, as well as the geographical and cultural distances between the interested parties, particularly for international registries. Simplicity depends upon having clear goals and adept leadership. Registry leadership may be in the form of an Advisory Board or other leadership committee. Advisory boards and general governance principles are discussed in Chapter 2, Section 2.6.

Rare disease registries present unique governance challenges because they often represent collaborations with many stakeholders and may be international in scope. Some examples are included below.

• Funding: If not centrally funded, who takes responsibility for raising money, writing grants, and securing the necessary funds for the operation of the registry? If fully funded, what role in decisionmaking does the financial sponsor(s) have? What are the startup costs versus maintenance costs?
• Privacy: Patients with rare diseases are more vulnerable than most to being identified by their health information. How does the registry protect the privacy of the individuals and families while at the same time creating a database of information and resources for the benefit of all persons having the rare disease of interest?
• Outreach: How does the registry identify affected and interested persons for participation in the registry?
• Information: What database is needed for the registry? What demographic, clinical, and/or longitudinal databases are needed? How will the registry adapt as new data needs are identified?
• Ownership: Who owns the information collected by the registry? For a tissue repository, who owns the materials in it?
• Agenda: Who sets priorities and establishes the work plan for the registry? What are the respective roles of the stakeholders in setting the agenda?
• Collaborations: Does the registry governing body create collaborations or can individual participants make collaborative agreements? What approvals are required? How are conflicts avoided and handled?
• Publications: Who takes responsibility for determining the publication plan, submitting abstracts, writing journal articles, and otherwise publishing about the activities of the registry? Who does the work? Who gets the credit? Are all of the professional and nonprofessional participants treated equally?