Data Donation Could Power the Learning Health Care System, Including Special Access Programs

In this issue, Walker and colleagues (Walker, Rogers, and Entwistle 2014) review the ethical dimensions of special access programs (SAPs) for unapproved medical interventions in advance of regulatory approval. Our own experience as advocates and researchers is with amyotrophic lateral sclerosis (ALS), a rapidly progressive and terminal condition with no effective treatment (Wijesekera and Leigh 2009) Over the past 15 years we built online forums for thousands of engaged ALS patients at BUILD-UK.net and ALS.net, and more recently a patient-powered research network (PPRN) at PatientsLikeMe.com (Brownstein et al. 2009). The PatientsLikeMe communities have nurtured many patient leaders across many diseases, but it is ALS patients who have most consistently advocated for earlier access to experimental treatments.

Patients find themselves in between two converging forces. On the positive side, “Moore's law” describes the doubling in the number of transistors on an integrated circuit every 2 years from 1970 through 2010 and has lead to a blossoming of ubiquitous consumer technology that (broadly) enhances their lives. On the negative side is “Eroom's law,” a semi-palindrome coined to describe the somewhat depressing halving in the number of new drugs approved by the FDA per billion dollars spent on R&D every 9 years since 1950 (Scannell et al. 2012). From the patient's point of view the frustratingly slow process of clinical experimentation clashes with the rapid pace of progress in supposedly less innovative spheres than medicine, such as home shopping, publishing, banking, or transportation.

We have seen the ethical tensions Walker and colleagues describe in ALS patients advocating fiercely for SAPs for a range of treatments, including prescription drugs (e.g., minocycline), novel therapies (e.g., dexpramipexole), and even experimental procedures (e.g., stem cell transplants). In parallel to these scientifically viable opportunities were complementary and alternative therapies ranging from pseudoscientific gadgetry to bee stings, which have started to be debunked by an international consortium of data sharers and truth seekers (Bedlack and Hardiman, 2009). In 2008 we witnessed patients conducting their own participant-led (Vayena 2013) or “apomediated” research (O’Connor 2013). When a small Phase II study from Italy suggested that lithium carbonate “delayed the progression of ALS” in 16 treated patients compared to 24 controls (Fornai et al., 2008), patients independently started taking lithium off-label under the supervision of clinicians who prescribed it on compassionate grounds rather than as part of a coordinated clinical trial. The potential benefits to patients seemed obvious; the Fornai and colleagues trial highlighted that in a 15-month period no patient taking lithium died, in contrast to 29% of the control group. In the face of such opportunity, who would pass up taking lithium now to wait for a clinical trial several years distant?

In a campaign that blurred the line between patient and researcher, patients and caregivers began encouraging those who had obtained lithium off-label to submit data about their dosage, weight, blood levels, and outcomes on the ALS functional rating scale (Cedarbaum et al. 1999) into a Google Spreadsheet, with the plan to analyze the data against historical controls (Frost et al. 2008). In response, we developed improvements on PatientsLikeMe that would strengthen data capture and invited this group to submit their data to our PPRN instead. Within 6 months of the Fornai publication more than 160 ALS patients were tracking their use of lithium on PatientsLikeMe. We used data from carefully matched historical controls in lieu of a placebo group and announced that we had failed to replicate the near-miraculous claimed effects of lithium within 9 months of the original study's publication (Wicks et al. 2008). We later published a more detailed analysis along with a complete deidentified copy of our data set and the code to replicate our matching algorithm (Wicks et al. 2011).

Initially the ALS research community was skeptical and criticism was raised that the rapid presentation of our findings made recruitment more challenging for the traditional studies. Even today, scientific reviews of the lithium carbonate story discount the patient-led study (Chiò and Mora 2013) even though its findings were replicated by the four subsequent randomized control trials. In hindsight, hundreds of patients were subjected to ultimately futile trials that blocked more promising treatments in the pipeline to “prove the answer” to a question that many patients felt had already been answered adequately by their efforts. Today patients continue to use the same tools to advance their own knowledge, even working to measure the effectiveness of drugs in formal clinical trials by tracking data and running participant-led research experiments on drugs, including those that would never be studied in the formal trial system and formal trials underway (Wicks, Vaughan, and Heywood 2014).

Despite promising beginnings, it is sad to say that every pivotal ALS trial has failed to show it can slow the disease when subjected to the gold standard of evidence, the double-blind placebo-controlled randomized controlled trial (RCT), with the exception of Riluzole, which may slow it by 2–4 months (Miller, Mitchell, and Moore 2012). More alarmingly, several of the experimental treatments trialed in the recent past such as minocycline (Gordon et al. 2007) and pentoxifylline (Meininger et al. 2006) appeared to hasten death and/or cause severe side effects that would have been hard to detect at the individual N of 1. In every trial of the past 15 years we have seen online evidence of patient lobbying for SAPs, and again with the benefit of hindsight we can see that SAPs would have provided no tangible benefits for either the individual or broader society (though intangible benefits such as hope or fighting the disease should not be underestimated).

Walker and colleagues ask whether it might be reasonable to make SAPs conditional on systematic data submission (“data-SAP”) by eligible patients. What differences might there have been, had a data-SAP been in place that was easily accessible to any ALS patient and was made a condition of early access? We can see a broad array of positives (e.g., more data, more patients involved, systematic data collection of what was previously happening unmonitored) but also many negatives (e.g., poorer quality data, less safety monitoring, more patients exposed to harms, medical resources being diverted to deal with adverse events, regulatory and legal uncertainty, erosion of recruitment from the formal clinical trials). Considering the range of treatments trialed in ALS, the risk/benefit analysis is very different for relatively benign interventions such as diet or supplements, as compared to prescription medication with known safety profiles being used off-label, as compared to wholly novel experimental treatments. While we do not support the furthest logical extension of this argument (anyone can try anything), we would certainly support the further building of infrastructure to allow such comparisons on a level playing field. The existence of organizations such as My Tomorrows, Empower Access to Medicine, Free to Choose Medicine, and the ALS Emergency Treatment Fund demonstrate that patients remain willing to mobilize.

Taking a broader view, the issues Walker and colleagues illuminate could be applied to the entire medical system. Every time an individual receives treatment but does not have his or her experience of efficacy, tolerability, and outcomes shared systematically is a loss to us all. If we’re going to build a system capable of learning from the experience of every patient on a given treatment and plotting it against sophisticated models of every other patient like them, why restrict such tools just to SAPs of new therapies? The promise of “big data” from the fields of consumer shopping, aviation safety, or telecommunications isn't just that these data sets contain a larger sample of a selected population, but that they contain data about every transaction, every sensor, every facet measurable of an operation continuously and in real-time. The nature of digital technologies is that a system fit for purpose in a data-SAP environment might well have applications in the tapestry of medical care data more generally, and that is a guiding principle of our blueprint for PatientsLikeMe (Weber, Mandl, and Kohane 2014). Fully realized, such a system could overcome some of the limitations of RCTs, such as multiple comorbidities, a lack of patient-centered outcome measures, long-term safety, or drug interactions that occur in the real world, while also allowing every patient to contribute to the fight against disease. At the core of any such program is good measurement, so that we can ensure that the system rewards those who are improving the outcomes that matter most to patients. This must be built on a framework of rigorous, openly licensed, adaptable measures if we are to construct a strong system (Harrington et al. 2014). Patients want to help contribute their data to help other patients like them, and its time we gathered the measures, the will, and the data to honor that.

ACKNOWLEDGMENTS

We thank the patient advocates Humberto Macedo (deceased), Rob Tison (deceased), and Eric Valor, as well as caregiver Karen Felzer, for their work in championing participant-led research in the ALS community.