Science and Technology CommitteeWritten evidence submitted by PatientsLikeMe UK

Submitted by: Paul Wicks, PhD., Managing Director. Dr Wicks is a neuropsychologist (Institute of Psychiatry, King’s College London) specializing in the conduct of clinical research using the Internet. He worked as a researcher for six years investigating cognitive aspects of motor neuron disease and later Parkinson’s at King’s College Hospital. For the past decade he has been involved in using the Internet to improve clinical trials, develop new measures of disease, and accelerate the pace of clinical research.

Disclosures: Paul Wicks is an employee of PatientsLikeMe and owns stock/stock options in the company. The PatientsLikeMe R&D team has received research support from Abbott, Acorda, the AKU Society, AstraZeneca, Avanir, Biogen, Genzyme, Johnson & Johnson, Merck, NIH, Novartis, Robert Wood Johnson Foundation, Sanofi, and UCB. PW sits on the advisory board of Current Controlled Trials for Biomed Central and is an associate editor at the Journal of Medical Internet Research.

Summary

1. A clinical trial is one of the most expensive, complicated, but scientifically robust tools available to medicine.

2. When the data arising from this exercise are compressed into a poster, conference presentation, or journal article, a great deal of information is lost that could benefit patients, clinicians, and researchers.

3. Specifically, the presentation only of averages and summary tables (rather than raw data) makes it difficult for future researchers to conduct meta-analyses ie to combine and summarize many studies to reach a more robust finding than could be reached by any one finding alone.

4. A mandatory open registry of machine-readable trial data could be constructed which would permit those who make healthcare decisions based on evidence to make better decisions.

5. Software developers could build applications that help patients to more easily interpret such data and make better shared decisions about their care with their doctors.

6. Such an undertaking might initially be considered burdensome or expensive but may represent a useful mechanism to make better decisions about the most effective treatments available given constrained resources in the healthcare system.

7. A level playing field for the presentation of evidence generated by clinical trials should reward those with the most effective products and so stimulate innovation.

Introduction

8. The Science and Technology Committee of the House of asked “How could the occurrence and results of clinical trials be made more open to scrutiny? Who should be responsible?”

9. In times of austerity we need to spend what resources we have on the best treatment options, but the status quo clouds and obfuscates data through publishing mechanisms rendered obsolete by the Internet.

10. PatientsLikeMe is an online patient network that allows those with long-term conditions to record and share their health data, connect with their peers, contribute to research, and identify clinical trials for which they might be eligible. Over the past seven years we have conducted studies investigating attitudes to trial participation.

11. We find that patients take part in clinical trials out of altruism, to help researchers make a better life for the next generation of patients to come. If we fail to make maximum use of the value generated by trials, we violate that social contract.

12. We believe that a repository of machine-readable clinical trial data, accessible to all free of charge, would be of major benefit to patients, researchers, health services, and government.

13. Through PatientsLikeMe we have demonstrated that with the right information visualizations, explanation, and importantly, the support of their peers, patients with no formal medical training can easily understand many of the important aspects of clinical data. After all, they are the ones living with disease and taking their treatments day after day.

14. Such systems lie within relatively easy reach, but as we have seen with ClinicalTrials.gov, successful implementation can only be driven by legislation.

Background

15. PatientsLikeMe UK is a wholly-owned European subsidiary of PatientsLikeMe, a company that provides a data platform for the exchange of information between users living with chronic illness. PatientsLikeMe provides a number of resources to its user base of nearly 200,000 patients, including an interface for the retrieval of clinical trials from www.clinicaltrials.gov.

16. PatientsLikeMe permits patients with long-term illnesses such as multiple sclerosis, epilepsy, or depression to complete validated patient-reported outcome measures alongside symptoms (eg pain, insomnia, fatigue), lab tests (eg white blood cell count, lung function), and treatments (including drugs, surgery, or behavioral modifications).

17. Members are able to share their experience of treatments and rate their perceived efficacy, tolerability, burden, side effects, adherence issues, and out-of-pocket costs, along with the opportunity to provide tips and advice to other patients who may be considering taking the treatment. The aim is not to provide medical advice per se, but rather to share the experience of what it is like to live well with illness.

18. This data has the limitation of not being collected as part of a double-blind randomized placebo-controlled clinical trial, the gold standard for evidence based medicine. Rather, the data provided on PatientsLikeMe reflects “real world evidence” from patients taking the treatment but assigned in a non-random way. Therefore the impressions voiced by patients are subject to bias such as the placebo effect, recall bias, or awareness only of perceptible side effects such as light-headedness but not imperceptible benefits such as stroke prevention.

19. Despite this, studies from our platform benefit from coming directly from the voice of the patient, being conducted rapidly, and investigating areas that would not normally receive research funding. To date we have published over 30 peer-reviewed scientific studies in clinical trial design, off-label drug prescribing, the development of new patient reported outcome measures, and barriers to medication adherence. We have conducted an observational study approaching the level of a clinical trial over the Internet to investigate the effects of lithium carbonate on motor neuron disease.

Problem:

20. Today, data collected in clinical trials benefits from being collected by physicians in a structured and carefully controlled manner. However, once a trial is completed, the data are locked away from public view. Dissemination channels have advantages and disadvantages:

21. Posters or oral presentations at conferences:

 

Rapid sharing of information with practitioners and researchers in the field

Presentations typically contain findings “hot off the press” including efficacy analyses and safety data and keep the scientific field informed of study progress

X

Results are summarized as a brief abstract (a multi-center trial of thousands of patients might be summarized in 250 words) or in briefly presented slides

X

While abstracts themselves may have been peer-reviewed before acceptance these are very concise and the final content will not have been peer-reviewed before being presentation

X

There is very limited availability of the findings outside conference attendees

22. As a published journal article in the peer-reviewed literature

 

Peer-review helps to ensure quality of science

Creates a citable document-of-record for future reference

Increasingly, modern journals such as “Biomed Central Trials” encourage the publication of trial protocols in advance of publishing the results which can then be cited to help ensure transparency and consistency throughout

X

Peer reviewers and readers can only see what the study authors have presented, typically summaries or averages. It is not possible for reviewers to check the statistical analyses from the raw data to identify error or obfuscation.

X

A single study is rarely enough to be conclusive. The best understanding comes from systematic meta-analysis of multiple studies conducted by different researchers—but data in a manuscript is rarely sufficient to allow meta-analysis

X

Journal access is generally restricted to expensive subscriptions, presenting a barrier for non-academic clinicians, private providers, patients, and those attempting to access information from the developing world. Contrast with the “open access” movement, for example

X

Negative studies are considerably harder to get published than positive studies, resulting in bias in the scientific literature

Opportunity

23. As a consumer, Which? or Amazon can instantly compare the profile of washing machines stratified by customer satisfaction, drum capacity, colour, price, or manufacturer, but the same level of granularity and information quality does not exist for patients to evaluate their treatment options.

24. The government, academia, and the pharmaceutical industry invest billions of pounds in conducting trials and generating evidence, yet the value inherent in this data remains unstructured, unsearchable, incomparable, and hidden from view.

25. We believe that society would benefit significantly from the transition of trial results from inaccessible records of unstructured text, tables, and graphs of the average to an open repository of machine-readable data to permit ongoing research, re-analysis, scrutiny, and meta-analysis. If we can do it with a washing machine, why not medicine?

26. Demonstration of an interim step along the way can be seen in ClinicalTrials.gov’s summary of trial data (enclosed). For instance, the record for NCT00355134, “Efficacy and Safety of Fingolimod (FTY720) in Patients With Relapsing-remitting Multiple Sclerosis (FREEDOMS II)” by Novartis® includes results from the completed study, publicly viewable online. Detailed data provided includes the number of participants recruited, their demographics, the number of patients lost to follow up, number of side effects reported, and the average group characteristics on the outcome measures of interest (which ultimately lead to the drug being approved as the first oral treatment for MS). This data was submitted by the manufacturer and is publicly viewable free of charge from the same record containing the trial protocol.

27. However, there is additional work to be done. Although open and well labeled, this report remains a document rather than a true dataset that could be re-analyzed by other researchers for purposes of assessing the natural history of MS, for instance. Whether the trial itself had been positive or negative, data from the placebo arm could contribute to a shared resource, which might allow future exploratory trials to be conducted without a placebo arm.

28. Additional benefits could be achieved by supporting such machine-readable data as part of structured informatics ontology such as ICF, ICD, SNOMED, and MEDDRA, such that data could be used across multiple diseases or allow more systematic interrogation of the data.

29. There is increasing recognition of the importance that patients become “engaged” with their own management, ie to take “actions (as) individuals … to obtain the greatest benefit from the health care services available to them”. “Engagement” stands in contrast to medical paternalism and advocates that patients take responsibility for their own healthcare in partnership with their healthcare professionals, such as understanding their disease, connecting with other patients like them, being compliant with their medications, and taking a proactive role in living well with illness.

30. For instance, patient engagement advocates promote “shared decision making” whereby the best decision is reached by the patient discussing and having a shared understanding of the risks, benefits, and alternatives of different treatments with their physician. For instance, a patient might find there are a dozen different medications available for their condition. While their doctor should be up to date on the latest literature the sheer volume of scientific research makes this practically impossible, and the scientific literature is inaccessible to patients.

31. By ensuring that product manufacturers submit machine-readable data arising from their Phase III and IV trials in a standardized format, it should be relatively simple for scientists, NHS staff, or entrepreneurs to create software programs that continually compare and contrast products on their risks and benefits. Additional data sources such as electronic medical records, patient safety registries, or extension studies could continue to increase users’ confidence levels by increasing the patient-exposure years reported through such a tool. Such a system could also permit patients who have taken (or considered) taking a given treatment to share their experiences from an anecdotal perspective, providing different levels of confidence for different classes of data provided.

32. Such a tool would also empower commissioners of NHS services to better make cost-effective decisions.

33. In addition to existing methods of dissemination (conferences and journal articles), archiving anonymised patient-level data in an open trials repository has the following advantages:

 

Permits re-analysis to identify subgroups, responders, or meta-analysis

Data can contribute to non-trial research such as natural history studies

Lessons learnt in terms of outcome responsiveness or population recruitment could accelerate future studies

Improve public confidence by assuring that doctors and researchers have full and unrestricted access to the data

A brief lay summary aimed at patients with the condition could help educate non-

Threats & Criticism

34. A recent debate in the European Parliament’s Environment and Health Committee on Clinical Trial Regulation (http://tacd-ip.org/archives/865, enclosed) details a number of objections put forward by opponents of trial transparency, which we will address.

35. Competition—In a world with open trial data and software solutions that help interpret this into information, manufacturers with a superior product should benefit by competing on efficacy and safety data, not marketing messages. While causing some short-term disruption to pharmaceutical manufacturers, open trial data will ultimately better reward innovative and more effective products. Embargo periods might also protect competition.

36. Administrative burden—Pharmaceutical companies have a robust and renowned capability to write regulatory filings in multiple territories with large volumes of documentation, isolate the prescribing habits of individual doctors from terabytes of data, identify fraudulent medicines in the farthest corner of the developing world, and to create high-quality scientific output en masse. They are therefore well equipped to deal with what is a comparatively small administrative burden.

37. Cost—It has been estimated that the cost of bringing a new treatment to market is between $300 million and $1.2 billion, depending on the study. In part this is due to a system that silos data and prevents stakeholders from learning from past errors. Barriers to entry force new researchers to learn how different outcomes measures will perform, which trial centres have the best recruitment, and the magnitude of placebo effects. An evenly distributed and enacted requirement for open trial data should reduce costs for manufacturers in the long term.

38. Our proposal will not be without cost and will require incentives to be adopted. We recommend that experts be consulted to provide a robust estimate of the costs of an open trial data repository.

39. No real need of greater openness—We would refer readers to Dr Ben Goldacre’s book “Bad Pharma” for a systematic argument against this notion.

40. Role of ethics committees and protection of human subjects—There are several potential approaches here, such as pseudo-anonymisation through adding random noise to the data. Ultimately no solution will be perfect and so we welcome proposals for legislation to further protect altruistic patients who share their data from risk or harms.

41. “Data is too complicated for the public to understand”. It should be noted that although the primary beneficiaries of open data will be other researchers, “the public” includes everybody, including those with scientific training necessary to correctly interpret scientific findings and patients themselves. Open and machine-readable trial data represents an opportunity for software developers to visualize and simplify the data into information, that is the entire point of information technology. Most people are not surveyors and yet Zoopla and MousePrice provide detailed information from the Land Registry to help them make better decisions about buying a home. Most people are not nutritionists and yet food labelling allows them to make better decisions. Most people are not cartographers and yet Google Maps helps them to navigate more clearly. Free the data, and complexity will be resolved.

42. Convenience—It may not be convenient to give the public open access to data on clinical trials, but initiatives such as 311 in the United States and Data.gov have shown that there can be substantial benefits. Furthermore the risks of large expenditures on ineffective treatments would seem to outweigh these. The major hurdles to overcome are cultural, not technical.

43. Permit the data to be used only by “independent experts”. Any restriction to open trial data would be harmful because “independent experts” do not exist. If a person has the experience and knowledge to design, run, or critique a trial, at some point they have likely worked for the pharmaceutical industry, received support from them, or are looking to do so in the future. There is no benefit in restricting the eyes that can access this data.

44. Potential for identification of participants with rare diseases—In our experience, patients with rare diseases are the ones most desperate to take part in research, and the siloed nature of the status quo makes participation difficult. For instance, recent reports (attached) in the Wall Street Journal of families with the rare disease Sanfilippo Syndrome show that parents resent having to take their children for multiple redundant blood draws and spinal taps in trials performed by different companies, when they have already provided similar data in other trials.

45. Consumer panic—Public trust in the pharmaceutical industry is low, and can only be strengthened by transparency, and although experts interpret data, it is ultimately patients who decide whether to ask their doctor for treatment, and once prescribed, whether or not to take it. Open data validated and certified by a trusted body such as NICE should reassure the public and rebuild trust in what is an innovative and important industry to the UK.

46. In conclusion we propose that the potential advantages of open, computable trial data far outweigh the disadvantages and will ultimately contribute to a continuously learning and improving healthcare system for the benefit of British citizens as well as industry and researchers.

February 2013

Prepared 16th September 2013