Advancing neuromuscular disease research through real-world data: Challenges and lessons learned

Commentary

Neuromuscular diseases (NMDs) are a diverse group of rare genetic disorders that affect the skeletal muscles, lower motor neurons, neuromuscular junctions or the nerves controlling muscle function. Individuals with NMD have significant healthcare utilization with associated costs ranging anywhere from approximately $3,000 to $500,000+ a year.^1–3 Efforts to mitigate unnecessary use and costs are important for patients, families, health systems, and payers, aligning with the institute of Healthcare Improvement’s Quadruple Aim.

Health administrative data are diagnostic, procedural, and billing information that are produced every time a patient interacts with the healthcare system. These data are ideal for performing epidemiological and healthcare utilization and longitudinal surveillance studies. This is particularly true in jurisdictions (countries/provinces/states/health networks) with single payer healthcare systems where data is centralized. There has been great interest in using health administrative data to study NMD as it allows longitudinal tracking of disease progression and clinical response to high-cost therapies to be evaluated outside of trial conditions. However, limitations of health administrative data include imprecise cohort identification and limited granularity of clinically important data. In the last decade the quality and number of NMD registries globally have improved dramatically.^4,5 The ability to integrate and link data from disease registries and health administrative datasets is an opportunity to overcome the limitations of health administrative data and generate real world evidence to answer clinical and policy questions.

This commentary will discuss important considerations and limitations associated with linking administrative data and NMD registries for real-world evidence generation. To do this, we will draw on our experience performing a multijurisdictional study that will link the Canadian Neuromuscular Disease Registry (CNDR) with health administrative data organizations nationally via Statistics Canada and in five Canadian provinces. In Canada each province administers its own unique health insurance plan that covers almost all medically necessary care in a decentralized manner. Provincial health administrative data contains information on almost all publicly funded healthcare with each province exercising data sovereignty. The CNDR (https://cndr.org/), since its inception in 2010, has recruited 7,000+ children and adults with NMDs from 43 clinics. Our study links CNDR cohorts of patients with Spinal Muscular Atrophy and Duchenne Muscular Dystrophy to health administrative data to better understand the disease’s natural history as well as the associated healthcare utilization and costs in the current era of disease modifying therapy.

There are notable challenges to performing this work which fall into three categories: linking registries to health administrative data, harmonization of data within and across jurisdictions, and the administrative burden associated with data linkage across jurisdictions.

Registries that collect data on individuals with NMD may not be designed to allow for linkage to health administrative data. Data sharing agreements between partner institutions that recruit participants and registries must include data linkage language and discussion of data custodianship. Furthermore, patient consent for the registry must also include explicit consent for linkage of their data to health administrative databases. If these processes are not in place, data sharing agreements will need to be revised, and patients will require re-consent.

The standardization of clinical data collection across NMD registries has been identified as a priority. When incorporating health administrative data there are additional data harmonization concerns. Within jurisdictions, data is frequently disaggregated across multiple databases. For example, data relevant to individuals with NMD may be housed in separate vital statistics, healthcare use, insurance claims, and prescription drug databases. In cases where healthcare use is funded by a mix of public and private payers or multiple private payers, combining these data sources may not be possible. This is especially relevant for NMD where costs are high, and patients frequently seek multiple funding sources. ⁶ Harmonization of data across jurisdictions can also be challenging because data is frequently not held in a consistent format from one organization to another. For example, each of the five Canadian provinces involved in this study have different data assets, documentation conventions, and formats. Outside of the Canadian context, this would be similar to combining data from different insurance payers or countries. This can be addressed using common data models, however, the infrastructure for the implementation of these models does not always exist and harmonization of data may be necessary to do on a project-by-project basis. How data will be harmonized and aggregated is important to consider during project conception as it will inform analytical planning.

Lastly unless pre-established legal and administrative processes exist, combining data across jurisdictions can be burdensome. In the Canadian context, health administrative data must remain within the province in which it originates. This is standard international practice as personal health data is typically not permitted to leave a jurisdiction. However, to aggregate outcomes and perform data analysis, personal health information in some capacity may need to be transferred in a deidentified/aggregated format to a central site. Given the decentralized nature of the Canadian context, transferring data between provinces is not dissimilar from transferring data between countries as each province has its own privacy legislation. Therefore, when performing this type of work, it is essential to establish which legislation governs the data and under what authority it may be used or transferred when performing multijurisdictional work.

Linking rare disease registries and health administrative data can be an invaluable strategy that allows the generation of real-world evidence to enable important and difficult questions about patients living with NMDs to be answered. Although there are challenges to performing this work, these can be overcome to advance NMD research. However, it is important to understand and plan for potential barriers as well as to advocate for infrastructure and legal improvements that will facilitate this type of research. Ensuring health administrative data linkage infrastructure is embedded within patient registries, the adoption of common data models, and consistent privacy legislation will facilitate future work of this type. Given the importance of efficiently generating real-world evidence for NMD in the era of disease modifying therapy, we hope that our experience and lessons learned can help prepare researchers in other jurisdictions who will take on similar work to plan and prepare.