NARCOMS and Other Registries in Multiple Sclerosis

Abstract

Observational studies and registries can play a critical role in elucidating the natural and treated history of multiple sclerosis (MS) and identifying factors associated with outcomes such as disability and health-related quality of life. The North American Research Committee on Multiple Sclerosis (NARCOMS) Registry is one of multiple registries worldwide that focuses on people with MS, but one of the very few patient-driven MS registries. On the 25th anniversary of the first data collection for the NARCOMS Registry, we discuss the importance of disease registries in the MS field, describe key concepts related to registry design and management, and highlight findings from MS registries relevant to clinical care or health policy.

Multiple sclerosis (MS) is a complex chronic disease, the etiology of which remains incompletely understood. Similar to disease etiology, factors that influence outcomes and many aspects of the MS disease experience are also incompletely understood. Observational studies and registries can play a critical role in elucidating these issues. The North American Research Committee on Multiple Sclerosis (NARCOMS) Registry is one of multiple registries worldwide that focus on people with MS,1 but one of the very few patient-driven MS registries. With operational support and funding from the Consortium of Multiple Sclerosis Centers (CMSC), the NARCOMS Registry project was initiated in 1993 under the leadership of Dr Timothy Vollmer. Data collection began in 1996 as part of the Yale MS Patient Registry and now is an ongoing project of the CMSC. Enrollment involves people with MS completing a questionnaire online or mailing it to the Registry Coordinating Center. Initially, data collection occurred only at enrollment, but longitudinal semiannual data collection began in January 2000. To date, more than 41,000 participants have enrolled in the NARCOMS Registry, contributing data to at least 140 peer-reviewed publications.

Herein, on the 25th anniversary of the first data collection for the NARCOMS Registry, we discuss the importance of disease registries in the MS field, describe key concepts related to registry design, and highlight findings from MS registries relevant to clinical care and health policy. We illustrate these concepts and findings using salient examples predominantly from the NARCOMS Registry. Characteristics of several MS registries are reviewed elsewhere.2

Why Are Registries Needed?

Several studies have highlighted the high burden of neurologic disorders worldwide. The Global Burden of Disease Study reported that in 2016 neurologic disorders were the leading cause of disability-adjusted life years, which reflect the combined effects of years of life lost and years lived with disability.3 That year, an estimated 18,932 deaths and 1,151,478 disability-adjusted life years were attributed to MS.4 Despite the adverse impact of MS and other neurologic disorders on affected individuals, their families, society, and health systems, critical gaps in our knowledge of these disorders persist. The 2020 Atlas of MS highlighted persistent gaps in knowledge regarding the incidence and prevalence of MS worldwide.5 The heterogeneity of MS outcomes, including treatment responses, remains incompletely understood.6 Clinical trials provide information about the effectiveness of therapies used under controlled conditions, but clinical trial populations often differ from populations managed in general clinical practice with respect to age, comorbidity burden, and other characteristics.7 Thus, the effectiveness, safety, and tolerability of therapies tested in clinical trials when applied in clinical populations are uncertain. Registries are one key way to fill these knowledge gaps to better inform clinical care and health policy. A systematic review of 17 studies found that clinical quality registries were associated with improved processes of care and outcomes.8

What Are Registries?

Registries are collections of information about individuals. They may be classified by their focus on 1) exposure to a particular disease or condition, such as MS; 2) exposure to a health care product, such as a device or drug; or 3) exposure to a particular health care service.9 Registries collect health status information using consistent observational (noninterventional) study approaches, although some registries may collect information after a specific intervention. Compared with other sources of real-world data, such as administrative databases or electronic medical records, registries have several strengths,10 including uniform data collection according to standard data definitions, the opportunity to collect long-term outcomes, and more comprehensive clinical data than may be obtained from administrative (health claims) data. They may also capture patient-reported outcomes (PROs). However, registries also face challenges related to sustainability, the potential for selection bias, and the need to monitor and maintain data quality.

The purpose of a registry should be defined explicitly before it is designed to ensure that the data collected will be appropriate for that purpose. Patient registries may be used to understand the natural and treated history of disease and prognostic factors; to understand disease treatment; to explore outcomes such as employment, socialization, and disparities in care; to measure quality of care; or to monitor treatment safety and harm.

The vision of the NARCOMS Registry is to improve clinical care and quality of life for persons with MS and their families through increased knowledge about MS. The mission of NARCOMS is to conduct epidemiological and health services research in MS; and to support new investigators and facilitate exploration of emerging research areas, as well as to participate in collaborative multicenter research focused in MS on its treatment and prevention. The goals of the NARCOMS Registry are to increase the understanding of MS from the perspective of the person with MS, to assess knowledge of treatment and health care services, and to disseminate knowledge and increase awareness of MS and its management to patients, their families, care providers, researchers, advocacy groups, and policy makers. The mission, vision, and goals of the NARCOMS Registry guide what data are collected and what projects are supported. As discussed further herein, the emphasis on the patient perspective underlies the focus on collecting data directly from the participant rather than from clinicians.

What Are the Important Design Issues for Registries?

Several critical issues should be considered when designing a registry. Legal and ethical issues may be particularly challenging. Presently, the NARCOMS Registry is approved by the institutional review board of Washington University in St Louis. Participants agree that their deidentified information may be used for research purposes and to be contacted regarding studies for which they may be eligible. Detailed discussions of legal and ethical issues can be found elsewhere.10,11 NARCOMS Registry participants agree to the use of their information for research purposes. Each survey receives institutional review board approval. Other important issues include the approach to recruitment and retention, data release procedures, and standard operating procedures; these are reviewed elsewhere.12,13 Depending on the purpose of the registry, the target population, the type of information to be collected, and the method of data collection will differ.

Target Population

The target population refers to the population to whom the registry findings are intended to be applied.14 A population-based registry aims to capture all potential participants in a given population, for example,15 all adults with MS in the United States. A population-based design is necessary for estimating the incidence of disease or for examining etiologic factors. A 2017 review of neurodegenerative disease registries highlighted several common sources of potential bias when registries are not population-based, such as inclusion of misdiagnosed cases, subtle changes in the types of cases included over time, failure to capture minority or disadvantaged populations, a skewed characterization of disease phenotype, as well as period and cohort effects.15

Registries that are not population-based can still address important questions, such as health outcomes, but assessing the generalizability of the registry population remains important to widespread applicability of the findings. The NARCOMS Registry is open to people with MS living anywhere in the world, although most participants reside in the United States. The Registry is not population-based. As part of the national MS Surveillance Registry feasibility project commissioned by the Agency for Toxic Substances and Disease Registry, the overlap was examined between participants in the NARCOMS Registry diagnosed before 2006 and persons captured in the Veterans Health Administration database, Veterans Benefits Administration database, and Medicare databases in Minnesota, Georgia, and South Carolina from 2001 to 2005.16 Overall, 28.8% of 1855 NARCOMS Registry participants were captured in the administrative data sets. However, 90.2% of NARCOMS Registry participants who reported having Medicare coverage were identified in that database. Similarly, 66% of NARCOMS Registry participants who reported that they were veterans were captured in the Veterans Health Administration or Veterans Benefits Administration database. Overall, one-quarter of NARCOMS Registry participants were not identified in any of the administrative data sources. This highlights the difficulty of capturing individuals with MS solely using administrative data in the United States and the consequent value of registries.

Data Collection Approaches

Registries may be broadly classified as clinician-driven or patient-driven. In a clinician-driven registry, the clinician is responsible for participant recruitment and data collection. A substantial proportion of MS registries are clinician-driven, including the New York State Multiple Sclerosis Consortium and the Veterans Affairs Multiple Sclerosis Surveillance Registry.17–19 Clinician-driven registries have several potential benefits. Engagement of clinicians enhances participant consent rates and recruitment of sufficient samples and helps ensure that participant diagnoses are accurate. However, these registries often lack the patient perspective, and data quality may suffer if the data collection burden on clinicians is high. The burden may be mitigated by integrating data collection with clinical care through electronic health records (see Box 1 for a description of the MS Surveillance Registry developed in the Veterans Health Administration).

Box 1. An example of an integrated registry in multiple sclerosis (MS)

What is the Multiple Sclerosis Surveillance Registry (MSSR)?

The MSSR is an MS registry focused on veterans with MS with the primary goal of optimizing their care.

How are the data collected?

• For each patient encounter, clinicians provide clinical information using the MS Assessment Tool, which can be accessed from the patient’s health record or from an online portal. This tool standardizes data collection and requires 5–15 minutes to complete.

• Data from other sources, including health care utilization files, pharmacy, prosthetics, laboratory, and radiology databases, are also integrated into the MSSR.

What are the implementation issues?

• Clinicians need to change workflow.

• Some information relevant to clinical care and research, such as neurologist-determined relapse occurrence and Expanded Disability Status Scale score, is not currently captured.

• A partnership between clinicians and information technology teams was needed.

^amaelstrom-research.org/mica/individual-study/mssr

In contrast, the NARCOMS Registry and iConquerMS are patient-driven registries. Patient-driven registries have the advantage of easily crossing geographic and jurisdictional boundaries20 and may be more cost-effective than clinician-driven registries. Selection bias is a potential concern due to the recruitment methods used and the inability to explicitly contact all eligible persons in the region or regions covered by the registry. For example, patient-driven registries typically rely on internet access and may miss individuals without such access or who use it infrequently; however recent studies suggest this may not be a major concern.21 Recruitment sources for the NARCOMS Registry include the CMSC and clinician offices, the National Multiple Sclerosis Society, the internet, NARCOMS Now magazine, and traditional media. The NARCOMS Registry offers recruitment materials to clinicians to support discussions about accessible opportunities to engage in meaningful research that has the potential to inform clinical practice. The addition of alternative methods of participation also mitigates potential biases in patient-driven registries. To address this issue, the NARCOMS Registry offers participants the option to complete questionnaires either online or on paper by registering at narcoms. org/participate. On average, 60% of participants opt to complete semiannual questionnaires online. It may be harder to retain participants long-term when there is no direct engagement of clinicians. For example, attrition is a greater problem in internet-delivered interventions than in face-to-face interventions.22 Concerns about data quality may also be raised because the data are not verified by a clinician or data manager at the site of interaction.

In 2005, recognizing the importance of the accuracy of MS diagnosis for the NARCOMS Registry, we compared self-reported diagnoses of Registry participants with physician-reported diagnoses and with diagnoses based on medical records review.23 We randomly sampled 240 participants, of whom 109 were considered to be active in the Registry and had current contact information. Of the 109 participants, 52 consented, 29 refused, and 28 did not respond, for a mean ± SD weighted response rate of 76.3% ± 4.5%. For each respondent we conducted interviews, requested medical records, and surveyed their treating physicians when possible. Based on these three modalities, we confirmed a diagnosis of MS in a mean ± SD of 98.7% ± 1.3% of respondents, supporting the validity of the diagnoses reported by NARCOMS Registry participants. In addition to requesting medical record review or confirmation of diagnosis by a treating clinician, other strategies to increase confidence in self-reported diagnoses include capturing additional information regarding medication history or disease characteristics23 and linkage (with consent) to other data sources, such as administrative health claims data.24

Types of Data Collected

Broadly, registries may collect clinician-reported data, performance-based data, and patient-reported data. Clinician-driven registries tend to capture the first two types of data, and patient-driven registries tend to capture the last. It is uncommon for all three types of data to be collected in a single registry, although the MS Partners Advancing Technology and Health Solutions (MS PATHS) study aims to do so.25 The MS PATHS is an initiative collecting clinician-reported, performance-based, magnetic resonance imaging, and patient-reported data at ten participating MS clinics in three countries. The data are used to facilitate clinical care and for research.

Clinician-reported data commonly include diagnosis, disability status as assessed on neurologic examination (eg, using the Expanded Disability Status Scale26), relapses, and use of disease-modifying therapy. Performance-based measures provide a standardized means of assessing functional abilities, distinct from what the clinician assesses or the perceived performance of the participant. The Timed 25-Foot Walk test is one such measure.27 Patient-reported data may encompass a broad range of information, including sociodemographic characteristics, health services use, and PROs (see Table 1 for the core data elements captured in the NARCOMS Registry, and see https://www.maelstrom-research.org/mica/individual-study/narcoms for access to detailed metadata).

Table 1.

Examples of data elements collected by NARCOMS Registry ^a

Category	Specific elements
Sociodemographic information	Date of birth Gender Race/ethnicity Annual household income Highest level of education reached Marital status Health insurance status Employment status
Clinical characteristics/outcomes	Age at MS symptom onset Age at MS diagnosis Disability/symptom impact (measured using Patient-Determined Disease Steps and SymptoMScreen28 depending on year) Clinical course Relapses in past 6 months Comorbid conditions Health-related quality of life (measured using RAND-12 and Health Utilities Index–Mark III depending on year)
Health behaviors	Smoking Physical activity Height and weight
Health care services	Use of emergency departments Visits to health care providers Hospitalizations
Treatment	Use of disease-modifying therapies (name, period of use) Use of symptomatic therapies (not on every survey)
Special topicsb	Preferences regarding physician assistance in dying Bone health Diet Health literacy Use of assistive devices Health information sources Bowel and bladder function Trigeminal neuralgia Visual functioning Vertigo

Abbreviation: NARCOMS, North American Research Committee on Multiple Sclerosis.

^aMore information may be found at maelstrom-research.org/mica/individual-study/narcoms.

^bSelected examples.

The PROs capture information that can be assessed only from the perspective of the patient, such as health-related quality of life, pain, and fatigue. As such, PROs are highly relevant to people with MS. By design, PROs are collected directly from the patient and do not involve interpretation by or the influence of any person other than the patient. More than 80 MS-specific PROs have been developed, and many generic PROs also have been used in studies of MS.29 The NARCOMS Registry routinely captures disability status and health-related quality of life.

What Are the Data Quality and Management Issues?

Good data quality is essential to the utility of a registry and to the validity of conclusions that can be drawn from the data. Multiple factors may contribute to poor data quality, and they may be broadly classified as systematic or random.30 Common systematic causes of data errors in disease registries include ambiguous data definitions, unclear data collection guidelines, poorly designed data collection instruments, programming errors, lack of a data dictionary available for those collecting data, and lack of a plan for quality improvement. Thus, achieving a high degree of data quality requires an ongoing consistent approach to data collection, with user-friendly data collection forms and standardized definitions that are readily accessible to investigators and data collectors.30 Implementation of automatic data validation rules at the time of data entry can also be helpful. Where applicable, the training and auditing of data collectors is critical. Data quality should be evaluated routinely, and a plan to correct inaccurate data if appropriate and fill in incomplete data should be constructed. MSBase, for example, has developed a standardized process to evaluate data quality, including an assessment of completeness, the proportion of variables with recorded values corresponding to their true range, consistency of the findings across variables, and believability.31 MSBase also assesses data density and generalizability of the study population with respect to the known epidemiology of MS.

The NARCOMS Registry has made extensive efforts to validate key data elements and measures or to use measures for which the validity and reliability are already established. For example, in a sample of 32 participants, we measured simple percentage agreement between patient- and physician-reported year of diagnosis.32 For year of diagnosis, agreement was exact for 17 of 32 participants (53.1%), within 1 year for 20 of 32 (62.5%), and within 2 years for 25 of 32 (78.1%). The test-retest reliability of participant-reported year of MS diagnosis was high, with an intraclass correlation coefficient of 0.99. Findings were similar for year of MS symptom onset (intraclass correlation coefficient = 0.90). Disability status is reported using Patient-Determined Disease Steps, an ordinal measure that correlates strongly with a physician-scored Expanded Disability Status Scale score, distance reached during the Six-Minute Walk Test, and 12-item Multiple Sclerosis Walking Scale score.33 Comorbidity is assessed using a validated questionnaire.34 For that questionnaire, agreement between self-reported comorbid conditions and medical records was high (κ > 0.82) for diabetes and hypertension; substantial (κ = 0.62–0.80) for hyperlipidemia, thyroid disease, glaucoma, and lung disease; and moderate (κ = 0.43–0.56) for osteoporosis, irritable bowel syndrome, migraine, depression, heart disease, and anxiety disorders.

Registry management is resource intensive, which can challenge sustainability. Registries require ongoing engagement by clinicians or participants (with MS), regular communication with stakeholders, data collection and management systems, a data management team, and financial support for these infrastructure costs.13 For example, the NARCOMS Registry distributes the NARCOMS Now magazine to Registry participants to maintain engagement and share research findings from the NARCOMS Registry to which participants have contributed as well as other information of potential interest. The NARCOMS Registry has a formal data coordinating center for data management, and data are collected and managed using REDCap (Research Electronic Data Capture),35 a secure Web-based application that can be compliant with the Health Insurance Portability and Accountability Act of 1996 (HIPAA) and the General Data Protection Regulation. The specifics of where and how the data will be stored should be established at the outset. Organizational, physical, and technological safeguards can be used to protect the confidentiality and security of the data collected.

Currently, the NARCOMS Registry is housed at Washington University in St Louis. Initially the NARCOMS Registry was established at Yale University, and during its 25-year history it has moved to Phoenix, Arizona; Denver, Colorado; and Birmingham, Alabama, before reaching its current location. Some registries are funded by a parent organization (eg, foundation or government), which requires continued commitment by the parent organization to provide funding. For example, the French Multiple Sclerosis Registry (Observatoire Français de la Sclérose en Plaques [OFSEP]; clinicaltrials.gov identifier NCT02889965) has primarily been funded by the French government.36 Others use access fees to offset the cost of the requested analysis and ongoing operational costs. Philanthropy can also be an important source of funding. The NARCOMS Registry uses a combination of these funding approaches. Specifically, the NARCOMS Registry receives partial funding of operational costs from the CMSC. Other funding comes from academic and nonacademic/industry researchers requesting access to existing deidentified data or requesting collection of new data (see https://www.narcoms.org/researchers-providers for access details).

How Have MS Registries Informed Our Understanding of MS?

Globally, MS registries have contributed substantially to our understanding of MS. Previous reviews have summarized contributions by the Danish MS Registry and MSBase.17,37 Broadly, registries have provided information regarding the incidence of MS and risk factors, characterized disease progression over time, and assessed the impact of disease-modifying therapy and the comparative effectiveness of different disease-modifying therapies.19,37,38 MSBase, for example, showed that the strongest predictor of postpartum relapses in MS is the presence of prepartum relapses.39 The Danish MS Registry showed that switching to highly effective therapy rather than moderately effective therapy after a relapse is associated with a subsequently lower annualized relapse rate.40

The NARCOMS Registry has also informed our understanding of MS. For example, lower socioeconomic status is associated with greater disability among persons with MS, and lower socioeconomic status accounts for some of the racial disparities in disease severity observed in MS.41 Vascular comorbidity, including hypertension, hyperlipidemia, disability, and heart disease, is also associated with accelerated disability progression.42 Diet is associated with symptom severity; that is, a healthy diet is associated with lower odds of reporting severe fatigue, depression, and pain.43

Future Directions: Enhancing the Impact of Registries

Registries have made an important contribution to our understanding of MS, but multiple registries exist worldwide with different goals. A key question is how to make the best use of these registries given the resources that are required to sustain them. In 2016, Bebo et al2 provided an overview of major registries and ongoing cohort studies in the MS field with the goal of making recommendations to enhance their impact. Those recommendations included 1) create a federated network of cohorts; 2) standardize data collection and management; 3) identify and prioritize research questions; 4) encourage collection of physician-reported outcomes and PROs; 5) encourage technological innovation; 6) develop a universal informed consent process; and 7) provide sustainable funding.

Subsequently, the International Advisory Committee on Clinical Trials in Multiple Sclerosis made recommendations to improve the quality of real-world evidence for informing our understanding of MS and its treatments. Some of the recommendations extend those made by Bebo et al2 by focusing on specific actions that would allow real-world data collected through registries and other data sources to be used and combined more effectively.10 These included developing a metadata catalog of existing cohorts and registries, developing standards to facilitate data exchange, developing guidelines to aid data harmonization in MS, and developing a toolkit that could support the development of high-quality registries and other observational studies by providing information regarding ethics and privacy issues, standard informed consent language, and standard operating procedures for data collection.

The MS Metadata Collective is a metadata catalog being developed for North American registries and cohort studies in MS with the support of Maelstrom Research (see maelstrom-research.org/network/msmdc).44 It includes information regarding the NARCOMS Registry and 12 other studies. Metadata catalogs are discussed in more detail by Geys et al in this issue of IJMSC.45 Further expansion of metadata catalogs can enhance the visibility of registries (ie, their findability) and enhance opportunities for collaborative and comparative work.

Although it is still in development, the Multiple Sclerosis Metadata Collective provides the opportunity to identify similarities and differences in the types of variables collected as to whether different instruments are used to measure the same underlying construct, such as pain or fatigue. Because multiple differences exist, data harmonization will be an important step to support collaborative work between registries. Harmonization can enhance the comparability of similar measures across studies to ease joint analysis.46 When data have already been collected, this process is referred to as retrospective harmonization, and the Maelstrom Research guidelines offer an approach to achieving it.47 The Big MS Data Network has assessed the feasibility of retrospective data harmonization across five MS registries, including MSBase, OFSEP, and the Swedish, Danish, and Italian MS registries.48 The NARCOMS Registry collaborated with the UK MS Register and the German MS Register to illustrate the feasibility of data harmonization across registries. Specifically, the three registries applied the Maelstrom Research guidelines to collaboratively evaluate employment status.49 The NARCOMS Registry also applied existing methods to illustrate the feasibility of developing crosswalks between different instruments that assess common underlying constructs. Specifically, the NARCOMS Registry developed a crosswalk between two instruments that measure health-related quality of life: the Health Utilities Index–Mark III version and the RAND-12.50

Individual registries cannot capture all information necessary to answer all research questions of interest. Linkage of registry data to other data sources is a means of reducing direct data collection burden, expanding the scope of questions that can be answered, and enhancing registry impact.24 Potentially linkable data sources include administrative (health claims) data, census data, and imaging and laboratory data. Multiple MS studies have linked registries with administrative health data.51,52 In 2015-2016, the NARCOMS Registry conducted a study in collaboration with IQVIA to test the feasibility of linking patient-reported data from the NARCOMS Registry to health claims data.53 Of 1000 registry participants approached, 769 (76.9%) consented to data linkage. Linkage to open medical claims and prescription claims data was achieved for 698 participants (91%) using IQVIA’s HIPAA-compliant encryption engine. These findings supported the feasibility of data linkage between the NARCOMS Registry and health claims data.

Conclusions

Registries offer an important means of enhancing our understanding of chronic diseases such as MS, which is a complex and heterogeneous disease. Multiple MS registries exist worldwide and have provided important insights into the epidemiology and outcomes of the disease, supported comparative effectiveness studies, and provided valuable patient perspectives. Efforts are growing to support collaborations across registries and data linkage to further enhance our understanding of MS.

PRACTICE POINTS

• Registries are collections of information about individuals. The information can be provided by the individual or a clinician, or it may be directly from electronic data sources.

• Important design issues for registries include the target population, type of information to be collected, and data collection approach (see examples at maelstrom-research.org/network/msmdc).

• Registries have informed the understanding of the epidemiology of MS and its treatments. For example, they have characterized disease progression over time and established the comparative effectiveness of different disease-modifying therapies.