Fresenius Medical Care, the leading provider of products and services treating end-stage kidney disease, has created the world’s largest longitudinal database of clinical dialysis information. When the initiative was still in the planning stages, the company knew it would need the expertise of a specialist to help safeguard the identity of the approximately 540,000 patients whose health data would be captured. That specialist was Privacy Analytics, whom Fresenius Medical Care had known through a prior project in association with the Renal Research Institute.
The Challenge
Health data from 40+ countries with differing privacy rules
Fresenius Medical Care is the leading provider of products and services treating end-stage renal disease. The company operates a global network of close to 4,000 dialysis clinics in more than 40 countries and is committed to advancing science. The company’s analytical teams throughout the world (e.g., analysts, data scientists) have long been using its clinical data in individual localities to assess care and understand the risk that various health concerns pose for people with chronic kidney failure. Their goal, however, was to create a single global dataset so their business, research, and quality improvement analytics could broadly and consistently be applied in healthcare analytics, including predictive modeling efforts that enable more personalized and precise treatments. Fresenius Medical Care recognized that a data set representing nearly a quarter of all dialysis patients in the world would be very valuable to both researchers and clinicians.
"Today approved analysts can get data on global populations in less than a week, a 25-fold increase in speed to get data needed compared to the prior state."
John Larkin,
Collaborations Lead with Fresenius Medical Care
There were two significant hurdles to aggregating the data across regions. The first was technical: the data was not harmonized, with each country or region using its own clinical systems and data fields. The second was privacy-related, as each area had its own privacy regulations and data-handling restrictions. John Larkin, Fresenius Medical Care’s Collaborations Lead, noted, “We could not move forward with a database like this without outside expertise on patient privacy. I don’t know many healthcare companies that could.”
The Solution
A unified plan for data transformation
When the initiative was still in the planning stages, Fresenius Medical Care knew it would need the expertise of a specialist to help safeguard the identity of the patients whose health data would be captured. The company turned to Privacy Analytics, whom Fresenius Medical Care had known through a prior project in association with its subsidiary the Renal Research Institute.
Once Fresenius Medical Care had completed the task of harmonizing the data across countries with a common data dictionary, Privacy Analytics went to work. Privacy Analytics first performed an assessment to determine the safety and security of the internal environment in which the data would be housed and used. The high marks that Fresenius Medical Care received on that assessment meant they could reduce the level of changes to the data during the anonymization process and could preserve more useful information in the fields most important to them.
Privacy Analytics was unable to get direct access to the full dataset to evaluate it, because parts of the data needed to stay in their country of origin. The senior data scientist from Privacy Analytics used an innovative privacy-preserving method, working from cross-sectional extracts of the data to conduct a re-identification risk determination and make recommendations on necessary changes to the data. These extracts revealed the distribution of personal information sufficiently to test various scenarios. Larkin said, “It was a genius approach and provided a way to get past a hurdle that we didn’t know how to overcome.”
Anonymize data to global standards
Regardless of the size and scale of your organization, we can help you make safe, lawful, and responsible use of sensitive data to improve services and create value with confidence. Download this fact sheet to learn more.
With that solved, Privacy Analytics devised an anonymization strategy that could be employed across all jurisdictions, even with a diverse range of populations and differing regulations. Privacy Analytics recommended a strategic set of changes to the data that would align with the most stringent privacy regulations, including the EU’s General Data Protection Regulation (GDPR) and others.
However, within that anonymization strategy, Privacy Analytics made every effort to customize the changes to the data to get the most useful and relevant information so that Fresenius Medical Care could unify and standardize analytics to drive insights that improve outcomes for patients. “We appreciated,” said Larkin, “that Privacy Analytics worked to squeeze every little drop of value out of the data. In order to retain certain elements that were most meaningful to us, such as patient weight, we had to make some concessions, but Privacy Analytics made it work.” Every such decision was weighed carefully through discussions with Fresenius Medical Care, and then carefully documented.
Larkin commented, “Privacy Analytics is the gold standard from our company’s point of view because of how well we can substantiate the reasons behind our data decisions.”
The Results
A resource to advance care
Fresenius Medical Care has now assembled its first global dialysis database coined Apollo Dial DB. The cloud-based data provides a highly sophisticated view into the clinical care provided to more than 540,000 dialysis patients. The data is fully anonymized and adheres to a complex set of global, regional, and local privacy requirements, including the U.S. Health Insurance Portability and Accountability Act (HIPAA) and the GDPR.
Importantly, Apollo Dial DB offers a streamlined pathway for analytics performed internally to drive the company’s research and development. Data on people with chronic kidney failure is now available to analysts and researchers more quickly and easily and is accessible anywhere in the world. Previously, it took about six months for analysts within Fresenius Medical Care to gather, process, and evaluate data, and the insights only pertained to one region. Today approved analysts can get data on global populations in less than a week, a 25-fold increase in speed to get data needed compared to the prior state. Apollo Dial DB thus significantly reduces the time needed to make valuable insights available to care providers. And in addition to this speed, analysts now have access to a large pool of patient data from around the world all at once, allowing them to perform studies that were not possible before and produce findings generalizable throughout the world.
Apollo Dial DB is quickly becoming a powerful tool for bringing insights and improving care quality and outcomes for patients. In fact, Fresenius Medical Care has already launched 15 clinical improvement projects based on the data. For example, one project is expanding the geographic scope of the Anemia Control Model (ACM), which is used to minimize the dose of anemia medication needed to maximize the hemoglobin levels of hemodialysis patients.1-3 In another project, researchers are characterizing medication patterns among dialysis patients across six continents to provide insights to clinicians.
Fresenius Medical Care foresees that Apollo Dial DB will become a resource to the broader medical community, with third parties able to use it to perform analytics under very specific project and data processing agreements. Such research will undoubtedly contribute to advancing the state of the art and provide a greater understanding of how to advance care for people with end-stage kidney disease.
1. Barbieri C, Molina M, Ponce P, et al. An international observational study suggests that artificial intelligence for clinical decision support optimizes anemia management in hemodialysis patients. Kidney Int 2016;90(2):422-429. DOI: 10.1016/j.kint.2016.03.036.
2. Bucalo ML, Barbieri C, Roca S, et al. The anaemia control model: Does it help nephrologists in therapeutic decision-making in the management of anaemia? Nefrologia (Engl Ed) 2018;38(5):491-502. DOI: 10.1016/j.nefro.2018.03.004.
3. Bellocchio F, Garbelli M, Apel C, et al. MO801: Use of the Anemia Control Model is Associated With Improved Hemoglobin Target Achievement, as Well as Lower Rates of Inappropriate ESA USE And Severe Anemia Among Dialysis Patients. Nephrology Dialysis Transplantation 2022;37(Supplement_3). DOI: 10.1093/ndt/gfac081.006.