Table 2.

List of recommendations for Improving the Quality of Rare Disease Registries.

Topics	Recommendations
Registry definition	#1 RD registries are organised information system, based on observational study designs with one or several predefined purposes and rules with a long-term perspective. We recommend that RD registries should not be limited by geographical area but have an international vision. In the field of RD, there is a need to identify as many patients as possible, and significant number of cases could be identified from a large population. Moreover, RD registries should be created and used as a data source for the purpose of performing subsequent or additional studies related to research, patient care or public health monitoring.
Registry classification	#2 In order to fulfil the expected objectives, the taxonomic position of the registry should be clearly defined. It can be classified into several categories according to different criteria (e.g., geographical coverage, diseases coverage, type of sponsors, methods used for data collection etc.).
Governance	#3 Define clear objectives and design the structure of the database consequently. Often registries start with a single purpose, but can evolve into multi-purposes, addressing the interests of other collaborators and stakeholders. In RD registries objectives should be expanded but it is preferable to prioritise and define them clearly. #4 Engage with all relevant stakeholders, especially patient representatives at an early stage in the implementation of a registry. #5 Establish a good registry team with clear role and responsibilities for all staff working members in proportion with the registry’s size, ambition and objectives. Identify a registry leader, domain expert, IT expert and ELSI expert from the start. Either employ an IT person within your registry or setup a long-term partnership with a commercial company. #6 Ensure compliance with (inter-) national and local ethical and legal requirements and on that basis develop public policies for accessing maintaining and operating the registry. #7 Ensure that the start-up required budgets have been evaluated and that the registry is well resourced for a pre-defined period. For long-term sustainability, registries should to seek funding from multiple, complementary sources.
Data Source	#8 The selection of primary data sources are critical for the success of the registry; they provide data of higher quality than secondary data sources; before incorporating a secondary data source into a registry, it is important to consider the legal and ethical feasibility of its incorporation and the potential impact of the data quality of the secondary data source on the overall data quality of the registry. Primary data sources are expensive and time consuming, so you should consider the possibility of reuse of existing data from secondary sources. #9 Provide clear-cut definition of the inclusion and non-inclusion criteria. In terms of geographical coverage, when selecting data sources (e.g., hospitals), it is important to consider the representativeness and ensure a satisfying level of quality in identifying all eligible patients (including the ability of proper follow-up).
Data Elements, Case Report Form, Standardisations	#10 The following steps are recommended for defining Data Elements: Determine what data are needed for the purpose(s) of the registry. Determine what information models and forms exist that can be reused. Determine what data are obtained from primary sources (requiring additional effort to collect) and from secondary sources (at the risk of lower data quality). Determine what data can be derived from other data, rather than being collected separately. Determine whether data can be collected and stored as part of routine clinical care (thus becoming data from secondary source). Determine whether data can be fed back to assist clinical care. #11 Ensure and promote the use of standards in the registry system (a) for diseases classification such as ORPHANET and RD Ontology (ORDO) and (b) for phenotypes description (such as The Human Phenotype Ontology (HPO). Use of standards facilitates the data interoperability.
IT Infrastructure complying with FAIR principles	#12 Involve registry users, the IT department and legal counsel in the selection process for the IT infrastructure. Select an off-the-shelf solution that is open and provides data exports in FAIR and common open data formats. Conduct a security assessment as part of the implementation track of your solution.
Data Quality	#13 In order to address data quality, introduce quality assurance and quality control activities at different levels. Monitor from the start of data collection and then regulatory data quality at central level and locally (site monitoring); produce regular data quality reports including evaluation of different dimensions of data quality (e.g., completeness, accuracy, duplicate prevention and timeless).
Quality information	#14 Develop a plan for statistical analysis describing the statistical techniques to be used in order to address the objective(s) of the registry; ensure data dissemination to different stakeholders: registry participants’, patients, general public, decision makers and researchers.
Documentation	#15 Developing and maintaining transparent and adequate documentation is essential for ensuring the quality and efficient operation of the registry. The detail of documentation may vary from registry to registry depending on the complexity of the registry.
Training	#16 Ensureproper and systematic training at all levels, addressed to registry staff and data providers. Provide training in a systematic way prior registry access by new users and when changes occur.
Data quality audit	#17 Have an audit system including defined triggers initializing audit processes.