Home / Resources / Blog / SNDS: A Goldmine for Effective Decision Making?

SNDS: A Goldmine for Effective Decision Making?

Real-World Data (RWD) have been increasingly used to further understand the natural history of a disease or condition, treatment patterns in RW practice, as well as the effectiveness, safety and costs associated with treatment options. They help exploring research questions, potentially complement clinical trial findings, and fill knowledge gaps. Different existing data sources can be used to generate evidence using RWD. The most well-known databases are probably the US health claims; in Europe, the French National administrative health data system, ‘SNDS’ is an emerging source for data.

What is the SNDS?

The SNDS — the French Healthcare database —  includes administrative information, primary care data, hospital data, and mortality data. As the largest and potentially most comprehensive healthcare data resource in Europe, it covers ~65 million lives, more than 99% of the French population and about 10% of Europe.

Until recently, French data were scattered among multiple databases, were difficult to access, and making links between them was not easy. Since the arrival of two key laws regulating the access to SNDS and the governance from the Health Data Hub (HDH), the SNDS has been made accessible to private companies, through an authorized third party. The authorization process involves a scientific committee verifying the feasibility, relevance and public health interest, and an assessment by the French data protection authority, CNIL.

RWD analysis requires both diversity and quantity in data. If several sources allow different questions to be asked, answering them involves having appropriate knowledge and awareness of the strengths and limitations of such RWD. Some typical challenges of working with administrative claims databases don’t apply to SNDS:

  • Representation is nearly total. US and European databases typically represent only a subset of the complete population. The universal French system covers all citizens, and the fact that most services are covered reduces biases: reimbursed claims represent the actual healthcare resource utilization (HRU).
  • Completeness: all primary care, hospital care and mortality data are available, for both in and outpatients.
  • Long-term follow up is possible, with up to 10 years longitudinal data, as, unlike other countries, people don’t change their health plans.
  • Data access is through well-established procedures and can be granted to both public and private entities, with the involvement of authorized sub parties.
  • The size of the database allows conducting studies on very specific populations, such as those suffering from rare diseases or presenting rare clinical outcomes.
  • Data linkage is possible and can be direct to other health records such as registries or biobanks.

Why and how to generate RW evidence

Evidence generation for value demonstration of novel drugs falls into four items: characterize the disease, evaluate its burden, characterise the effect (therapeutic or adverse), and evaluate the impact of treatment. They address the requirements of different stakeholders: HTA/payers, regulators, drug developers, finally benefiting to patients, informing healthcare practitioners, and helping providers and payers in decision-making.

HTAs play an important role in determining the added value of a given health technology compared to existing ones, assessing them from clinical and economic perspectives that eventually facilitates decisions regarding the pricing, reimbursement, and access to patients. Regulators also use RWD to monitor for and act on any unforeseen risks with medicines following marketing approval. From a strategic standpoint, RWD allow depicting an overall picture of actual treatment practices and healthcare delivery, providing the best information to make the right decisions for patients. With its comprehensiveness and robustness, the SNDS can help fulfil most of the requirements and is perfectly suitable for gap analysis, identification and quantification of unmet needs, description of treatment patterns, and characterization of a disease’s clinical and economic burden.

Even if the market access landscape in Europe and in the US is complex and differs from country to country, there are many similarities in how HTAs typically assess a health technology. Furthermore, while evidence generated using SNDS data reflects clinical practice in France and per se is the most useful for the French landscape, RWE generated in one country is still impactful for decision making in others, in terms of disease understanding, clinical and economic burden evaluation, and pricing.

What’s next?

Whilst the SNDS is currently like other global health claims databases, in terms of data available, it offers the ability to conduct robust, generalizable research on most diseases (including rare disorders) with long-term follow up. Recent case studies have leveraged many of these advantages, including for HTAs and market access. Other advantages will be uncovered as the research with the SNDS increases.

Despite of its strengths, the SNDS should not be considered a one-size-fits-all approach. Such RWD can be used in different contexts and for different stakeholders, allowing value demonstration and access. But it should be only an element of a global toolbox.

As for the other RWD, study design is key. RWD are only as good as the data source and methodology used to analyse that data. RWD analyses shouldn’t be used as stand-alone evidence, and results need cautious interpretation. Drawing reliable conclusions requires adequate and robust analyses, including advanced statistical methods designed to address potential issues, such as confounding factors. In addition, observational RW studies can only evaluate associations and not causality regarding the efficacy and/or safety of a treatment. The reliable use of such comprehensive data opens new perspectives, such as the development of bridging from clinical trials to RW modeling tools which allow for optimal study design. Again, cautious model qualification and awareness of the limitations and application domains are required.

A limitation of SNDS is that there can be inherent biases as an administrative claims database relying on coding. In addition, even if SNDS is increasingly being used and working practices have improved dramatically, including the development and validation of algorithms, some key clinical variables are still missing.

To complement the SNDS, many initiatives are ongoing, including the willingness to facilitate linkages to different relevant health records and to share more and more data as per the French HDH catalogue. Such additional data, collected from medical devices or imaging, registries, cohort data, health status tracking, will increase again the value of the SNDS database and the applications of these data-driven studies. In recent years, we’ve seen a massive growth in the amount of RW data available and the development of tools to study and gain value from ‘Big Data’, such as machine learning and artificial intelligence methods. Taken together – and in compliance with data privacy regulations– this information provides insights and touchpoints that could lead to better medicines and methods and improved patient care.

RWD challenges have changed, from the need for data to the need to develop and improve methods to analyse this data, taking into account the importance of data privacy and security. Recent initiatives from the EMA and the US FDA highlight the growing interest in RWE, beyond just safety evaluations, to helping define new subgroups of patients, new label indications, and designing clinical trials. 

So, is SNDS the gold standard to generate evidence for decision making? Certainly, it has important strengths and is appropriate for some objectives. But it must be thought of as part of an integrated strategy considering possible biases and limitations. The upcoming databases per the HDH and initiatives involving advanced modeling, machine learning technologies and artificial intelligence will enrich it and extend its scope of applications, which could benefit all stakeholders in the healthcare ecosystem. To learn more about how we’re using SNDS to generate evidence to quantify the opportunity for a new drug and justify market access, click here:

About the author

Nadia Quignot
By: Nadia Quignot
Nadia Quignot is Director, Real World Data solutions & Data analytics at Certara. Over the last >12 years, Nadia has worked on the development of methods to inform risk and benefit/risk assessment and has coordinated a range of (real world) data analytics projects, for European bodies and pharma companies. As a safety scientist at Roche Pharma, she was in charge of risk and benefit/risk evaluations for anti-inflammatory and anti-cancer drugs. In the R&D projects she conducted for European bodies, she was particularly involved in hazard characterization of drugs, pesticides and other food-related and environmental contaminants. Nadia joined Analytica Laser in 2013 and has developed an extensive expertise in real-world evidence strategies and model-based analytics. She has been coordinating modelling initiatives and a range of (real world) data analytics projects, for European bodies and pharma companies, including directing >15 French SNDS database projects. Nadia holds a PhD in Pharmaco/toxicometrics from Paris 5 René Descartes University.