In rare disease, every challenge is magnified, beginning with the first. Simply put, identifying a disorder and receiving a formal diagnosis is at the root of myriad difficulties—from connecting a patient to a community to generating an appropriate care and treatment plan that ideally goes beyond managing symptoms to connecting a patient to clinical trials.
Related challenges exist for other healthcare stakeholders. Can a system deliver effective treatments when it cannot identify an illness? Can life sciences companies raise funds and develop treatments without a proper idea of how many patients they may eventually treat, or what the patient experience with the disease is? Can insurers establish their burden of disease if it is difficult to identify?
These questions raise many implications, but advancements in healthcare data and analytics can make a huge impact on patients and healthcare providers, including reducing time to diagnosis and increasing connectivity for patients to the right providers, well-matched trials, and patient communities.
The State of Healthcare Data for Rare Disease Patients
Access to basic healthcare data across the life sciences industry is table stakes for research and development, clinical development, and commercialization. For more than a decade, access to deidentified sources of real-world data—especially administrative claims, patient registries, and more recently electronic health records (EHRs)—has been informing unmet needs, population sizing and forecasting, and commercialization and marketing activities.
However, data is always missing, whether because of source coverage (e.g., a particular health system’s data is absent), or because data was never created in the first place (in particular, if a doctor does not or cannot register a formal diagnosis, then a disease may fly under the radar in even the best datasets). While true of most diseases and data, the impact on rare diseases is significant.
An emerging technique for overcoming this challenge is to apply a combination of clinical expertise with predictive analytics to identify likely patients.
Applying Predictive Analytics to Identify Probable Patients
The Orphan Drug Act defines rare diseases as those affecting fewer than 200,000 people in the U.S. It’s very common to look at datasets claiming to have information on more than 300 million Americans and estimate a rare disease prevalence that is dramatically lower than the best estimates otherwise available. But this makes sense when you consider that if a rare disease takes five years to diagnose, then that’s five years in which it won’t appear in medical data.
However, in the same way marketing professionals can use age, gender, and online activity to predict what smartphone you will buy, analytical techniques can be combined with clinical knowledge to predict that a patient is likely to have an illness based on the symptoms presented in their medical history or EHRs. It may be unreasonable to expect a doctor to remember the diagnostic criteria for each of 7,000+ rare diseases, but it’s quite reasonable to ask a machine to do that.
Once identified as likely, it’s much easier to engage a patient’s physician to conduct testing, refer to centers of excellence, or recommend clinical trial enrollment. The first and arguably most important activity to help rare disease patients is to help them receive an accurate diagnosis.
Just as this usage of data can help healthcare stakeholders target the right patients to engage and guide on a successful path to diagnosis, this same data is being deployed in different contexts to reduce delays in R&D and clinical developments.
Rare Disease and Real-world Evidence
Governments around the world are implementing guidelines and developing strategies for the role real-world data (RWD) will play in regulatory decision-making. But the main question is how we can use health data generated from numerous sources—those named above as well as novel sources such as smartwatches, sensors, and medical devices—to complement or supplement controlled clinical trials. If we can, then imagine the cost and time savings or the speed to treatment.
The implications for rare diseases are obvious. How can we design clinical trials without an understanding of the natural history of disease? How can we recruit patients for trials if they aren’t already diagnosed? Even if we can find enough patients, it is not possible, or ethical, to assign patients to a placebo arm because doing so deprives one of the so few patients of the potential treatment’s benefits and reduces the amount of valuable evidence that can inform regulatory decision-making.
Researchers and institutions are studying how RWD might benefit rare disease clinical development, and the number of registries and partnerships in this area is growing. This revolution in the application of data and analytics to increase healthcare effectiveness and to increase clinical trial participation for rare disease patients requires sufficient data.
Patient Data: From Patients, Of Patients, For Patients
Historically, this type of RWD is primarily generated through the provision of care or else through purpose-built registries. These sources come with their own difficulties: What data is missing? How easily can new data be connected to improve our understanding of disease and the patient experience? How long will it take to accrue sufficient patients to generate insights? If this data is accessed, is it possible to engage patients directly to learn more, to collect missing information, and to generate evidence prospectively?
In the U.S., the 21st Century Cures Act unleashed a set of requirements for healthcare entities to enable more frictionless data sharing for certain purposes, which in turn are spurning innovation, venture activity, and numerous companies to build the pipes to enable it.1 This means that in 2023 and beyond it will be much more conceivable that a community of rare disease patients could opt their data into studies and research that may then be used to improve care.
Companies that characterize themselves as “patient data lockers,” health information exchanges, and similar are using emerging policies to enable data extraction directly from EHRs around the country at the behest of patients or appropriate parties.
Where it may have taken months to get access to one’s own data from multiple healthcare providers (if they accessed them at all), in principle this can now occur in minutes. Having seen this technology in action, it’s a staggering advance. Of course, this benefits all patients, but if you’re a pharmaceutical company struggling to find enough patient data to inform your development and commercialization activities, a clearer path to engaging patients directly is emerging and seeing results immediately.
The next frontier of drug development and commercialization is in rare diseases. To be effective, the use of data and analytics needs to break out of traditional molds, and the patients themselves must play a bigger role.