17 May 2011. Having transformed the way people interact, social networking may be poised to transform medicine as well. The Internet has already opened up new ways for people with chronic diseases to overcome isolation and to connect with others like themselves on social forums. With the advent of online patient registries, the trend is moving beyond discussion of symptoms and treatments to data sharing and online research studies published in leading scientific journals. One example of the latter appeared in the April 24 online Nature Biotechnology, authored by a pioneering group in this area. Is this trend a good thing? Some researchers express qualms about the quality of data gathered in this way, but most applaud the overall movement to develop registries for various diseases. Patient registries, sources said, have the potential to be a tremendous resource that will yield benefits to clinicians, researchers, and patients. They empower patients, speed recruitment for clinical trials, deepen data collection, and improve disease management, thanks to feedback from patients. All sources interviewed for this story agreed on another thing: Registries are the wave of the future.
Patient registries come in different flavors. One type is maintained by research clinics, where clinicians enter data and patients typically cannot access it. At the opposite end of the spectrum, many patient advocacy groups are establishing online registries where patients or their caregivers enter their own data. Some of these sites have open data, meaning a given person’s data is available to all members of the community. Can such data accelerate clinical discovery on tough diseases where progress feels woefully slow? The consensus at this early stage is that such data cannot replace clinical trials, which remain the gold standard for testing new treatments. Some researchers believe, however, that patient-supplied data may complement trial data, for example, by providing preliminary findings on safety or efficacy that may influence trial design, or by testing alternative treatments or off-label drug use that might never find a sponsor willing to pay for a standard, costly clinical trial. For its part, the Food and Drug Administration encourages trial sponsors to include patient-reported outcomes in trials.
The Nature Biotechnology paper describes one test of this approach. Senior author James Heywood, along with his brother Ben, co-founded the data-sharing platform PatientsLikeMe, Cambridge, Massachusetts. The free site allows people with various health conditions to share their experiences with others who have similar conditions. The company’s research team used data volunteered by patients on the site to evaluate the efficacy of off-label lithium use for amyotrophic lateral sclerosis (ALS) (see ARF related news story). Because in 2008 a small clinical study suggested that lithium might slow disease progression, some ALS patients decided on their own to try the drug. PatientsLikeMe built a data collection tool where participants could enter information about their symptoms, treatment, and outcomes, using the Revised Functional Rating Scale to measure disease progression. After 12 months, 149 people who took the drug met inclusion and exclusion criteria for analysis.
The researchers compared patients who took the drug to controls who did not. To reduce potential bias in this observational study, PatientsLikeMe developed an algorithm to match patients to controls who resembled them in disease progression and characteristics, using historical data gathered by the site. For example, one potential bias would be if people who chose to self-medicate were also those who had the most severe or aggressive disease, said first author Paul Wicks, who is the R&D director at PatientsLikeMe. The authors controlled for this by matching up people who had similar disease severities and patterns of progression. The analysis showed no significant treatment effect from lithium, Wicks and colleagues report. This finding agrees with the results of later, larger—and more expensive—formal clinical trials of this drug for ALS (see ARF related news story and ARF news story), suggesting the paradigm produced valid results.
Lucie Bruijn, who leads the ALS Association’s research enterprise, cautioned that an observational study like this “wouldn’t be able to find small effects, and wouldn’t necessarily be able to find all the toxicities. One should be skeptical in terms of getting good, solid clinical data from this.” Bruijn noted that the approved ALS drug riluzole probably would not have been identified in this sort of study. Nonetheless, she praised the value of patient registries and suggested clinicians should work with registries to encourage people to enroll in clinical trials. “It’s a great tool to engage the people with the disease.”
On the other hand, data expert Jeff Hammerbacher said that all data are useful; observational studies simply require different analytical tools. Hammerbacher is the founder of the software company Cloudera, San Francisco, California, and a director at Sage Bionetworks, Seattle, Washington, a nonprofit foundation that creates global disease datasets. Prior to Cloudera, Hammerbacher oversaw the team that built the core data platform at Facebook; he is considered a leader in cloud computing and crowdsourcing, two computer science buzzwords that refer to managing vast databases across networked computers, and recruiting large groups of volunteers to work on projects through the Web. Hammerbacher said that large patient registries offer a greater depth of data per patient, a longer timeframe of data collection, and often a larger number of participants, making these data extremely valuable.
Indeed, most people interviewed for this story agreed that patient registries can improve data collection in many ways. For one thing, only a small fraction of people with a given disorder ever go to a research clinic. Online registries can potentially reach a much broader spectrum of patients nationwide. For another, registries can capture more varied data than clinical records do, such as lifestyle factors and day-to-day disease management, said Marie Schiller at Health Advances, headquartered in Weston, Massachusetts. Only a sliver of that rich phenotypic information is typically part of a clinical record, Schiller said. Registries could also enable epidemiological and longitudinal studies as they gather data over long time periods. When registry participants take part in clinical trials, researchers involved in the trial will be more easily able to contact them to see how they are doing months and years later, Schiller noted. The kind of detailed data gathered in a registry could help usher in the much-heralded era of personalized medicine, Schiller said, by identifying patients who could benefit from specific treatments.
Patient registries offer opportunities for quality improvement, Bruce Marshall, vice president of clinical affairs at the Cystic Fibrosis Foundation, wrote to ARF. Registry data can expose variations in treatment practices and clinical outcomes, and the databases could feed that data back to research centers to improve care for all patients. Patient registries empower people with disorders, Wicks said. Users can connect to other people with the same disorder, form support networks, ask questions. Frequently, newly diagnosed patients find it helpful to “get up to speed” with the advice of others who have fought the same disease for years. Schiller suggested that registry participants could also provide feedback on medical devices, letting companies know what they like and do not like about the design.
Wicks argues that studies by registries such as PatientsLikeMe can complement clinical trials. They can provide preliminary safety data, or tell researchers the approximate effect size for a drug, allowing scientists to better decide what sample size they need to achieve statistical power in a larger clinical trial, Wicks said. Registries can gather data more quickly and cheaply than conventional trials can. That makes it feasible to study treatments that are not profitable enough for trials, such as off-label uses of drugs, off-patent drugs, lifestyle modifications, or health food supplements or vitamins. Schiller noted that registries can collect data on trends, which, in turn, can generate hypotheses and inspire future trials. “I do believe that [registries] can at least foster ideas that can get translated into well-defined, objective-driven studies,” she said.
There are drawbacks to using registry data as well. Patients’ self-reports may be inaccurate, or the registry population may be biased. Data may be less standardized and noisier than trial data. Some diseases will be more suitable for registry use in research than others. In some disorders, notably advanced Alzheimer’s, patients cannot self-report, though caregivers could enter the information. Registries also have to deal with privacy concerns and safeguard sensitive medical information appropriately. Perhaps most seriously, some patients might choose to self-experiment with risky drugs or doses, something registry managers would like to discourage.
Whatever the risks, patient registries are here and growing, people contacted for this story agreed. They come in many forms. PatientsLikeMe is unique among patient registries in that it serves multiple conditions (see ARF related news story). In April 2011, the site opened its doors to any patient with any medical condition (see PatientsLikeMe video), and currently includes over 100,000 patients with more than 500 different conditions. PatientsLikeMe is also one of the most open sites, with all data freely visible to other users. Other groups, for example, Genetic Alliance, a nonprofit organization that encourages people with genetic disorders to get involved in advocacy efforts, also use the patient-driven model. Genetic Alliance allows people to complete a detailed family history that they can share with their doctors, but the site does not include a shared patient database. At the other end of the spectrum is the Cystic Fibrosis Foundation’s registry, a closed model in which only clinicians enter data. Founded more than 40 years ago, the registry includes data from almost 30,000 patients across the country, and is credited with helping to double the life expectancy of cystic fibrosis patients through improved treatments (see The New York Times story). “We are interested in patient-reported outcomes, and that may evolve toward data entry by patients or caregivers, but we see that as complementary to the medical data input by the health care professionals,” Marshall wrote to ARF.
A registry that is currently under development plans to bridge these extremes. Schiller is working with the Type I Diabetes (T1D) Program funded by the Leona M. & Harry B. Helmsley Charitable Trust to establish a national online and mobile platform registry, the T1D Exchange (see also story from Partnering for Cures meeting). They are currently enrolling participants, Schiller said, hoping to launch in the fall of 2011. Schiller sees this registry as a “third-generation” model that links the tried-and-true clinic-based registries with patient registries and fosters communication between the two groups. The T1D Exchange will incorporate data collected at type 1 diabetes research centers nationwide, and will also include a biorepository of patient blood and CSF samples. Comparing clinic data to patient-entered data will provide built-in quality control, Schiller noted. The program’s goal is to improve patients’ health as well as to advance research. “Clinicians are excited about the opportunities they see with the patient portal,” Schiller said. “They’re driving studies into that portal.”
Other organizations have not yet developed a patient registry, but are interested in the potential. Monica Coenraads, who runs the Rett Syndrome Research Trust, a nonprofit group that catalyzes research leading toward clinical trials, said there would be barriers to adapting a patient registry to this neurodevelopmental disorder. Not only can patients not self-report, but the wide variations both between patients and within a given patient over time would make data comparison difficult. Nonetheless, Coenraads said a registry might be helpful in many ways, such as looking at off-label drug use that might ameliorate some of the complications of the condition. It could also educate caregivers and reach out to people who find it difficult to leave their homes. She also thinks it is important to send information from patients back to clinicians and researchers. “There’s a tremendous amount to learn from patients and their families,” Coenraads said. “Communication should be a two-way street.”
For her part, Schiller envisions a future where registry participation is a matter of course for those with a chronic or progressive disease. “When you get diagnosed, you would get your first prescription and your key to enroll in the registry, and it would just become part of chronic disease management,” she said.—Madolyn Bowman Rogers.
Wicks P, Vaughan TE, Massagli MP, Heywood J. Accelerated clinical discovery using self-reported patient data collected online and a patient-matching algorithm. Nat Biotechnol. 2011 Apr 24. Abstract