Data sharing is one of the most important current issues in clinical trials. If conducted in the right way, sharing data could help facilitate faster and better generation of evidence to bring new interventions and improve disease management.
Many leading organisations involved in healthcare research recognise that we need international guidelines that would ensure data sharing can be done in such a way that maintains the integrity of the research, whilst also protecting the rights of research participants and the interests of the researchers.
To address this the Institute of Medicine have formed a committee and seek to gather wide perspectives with the aim of concluding a set of recommendations for data sharing in clinical trials.
It is essential that these recommendations are appropriate to all type of trials, are globally applicable and can be practically implemented.
The first report from this committee sets out this objective in detail and seeks input from as many different groups as possible.
This is a very important opportunity to contribute to the outcomes of this committee and influence these recommendations
These recommendations may well be adopted by research funding agencies, regulatory authorities, pharmaceutical companies and publishers. Therefore the committee are very keen to receive input, feedback, opinions, ideas and comments in order to be certain that the recommendations that this committee conclude are truly appropriate for every conceivable type of trial conducted anywhere is the world.
The purpose of this blog is to disseminate this report as widely as possible through The Global Health Network community (please forward the link to this page to everyone that you can think of), and we urge you to comment. Please simply write your response in the reply section below. This is important for three reasons:
- It is essential that the perspective, concerns and realities of conducting trials in low-resource settings is a key consideration of this committee and that the recommendations will be globally applicable and appropriate.
- If you write a comment here this will encourage others do so.
- These comments can be collated and I can present them to the committee in February. Obviously the more responses there are the more powerful the impact will be.
This is a very important opportunity to help the committee understand the settings and context in which we all work and what you think about data sharing:
- Could data sharing be beneficial in your field?
- What concerns do you have? What are the barriers?
- What could be done to ensure fair and appropriate data sharing?
If appropriate mechanisms can be found then data sharing could bring vast benefits to populations in low resources settings. Lets make sure those mechanisms are workable and will improve evidence-led improvement in global health.
You can download the report by clicking here.
We look forward to hearing from you all!
We are very pleased to be able to share two documents from Francois Houyez, Director of Treatment Information and Access at Eurordis.
You can access these documents by clicking on the titles below:
Transparency and Access to Documents - what it means to patients
A paper recently published on PloS Medicine [Karunakara U (2013) Data Sharing in a Humanitarian Organization: The Experience of Me´decins Sans
Frontie` res. PLoS Med 10(12): e1001562. doi:10.1371/journal.pmed.1001562] may help to discuss/address some of these crucial questions.
In the paper, the humanitarian organization MSF describes how its data sharing policy for routinely collected clinical and research data was developed; the principles underlying it; and the practical measures taken to facilitate data sharing.
Experiences in sharing clinical datas between different research groups and different countries may have an unsuccessful end due to the lack of standardization of data recording, clinical perception, and data processing.
To avoid this, the prospective studies with standardization of the collection, interpretation, and analysis process are indicated.
Collecting data in developing countries is hard and expensive. Researchers who do this are poorly paid and lack resources to develop the data to its maximum.
Data sharing should not be for free, it should recognise the power imbalance between those with the resources to analyse the data, and those who collect the data.
As such data sharing needs to be a partnership, whereby data are analysed in collaboration with those who collect the data. The more complex the analysis, then the more work needs to be done by the analysts to ensure that the researchers are able to understand and replicate the analysis.
Data sharing is not, and should never be seen as zero cost to either side of the data sharing partnership. But should be seen as an investment opportunity, with the greatest rewards coming to those that invest the most in the data sharing.
As much as there are challenges with data collection in low resource areas, the fact that low resource areas suffer impoverishness, communication barriers, illiteracy, poverty and the like, while designs inculcate all this into the project in My opinion I suggest that standalone easy to use computer gadgets as little as mobile fones, be used to collect this data onsite. This systems will require the user (locals trained on it) to follow step by step guide to collect this data.
We should incorporate the technology of GPS system in data collection, robots should be inculcated in design implementation for data collection in both in low and high resources areas. This will go a long way to reduce discrepancies as regards data collection in the areas.
E.g you take a picture, GPS TAGS THE LOCATION OF THE PICTURE AS IT IS POSTED ONLINE.
Theres the challenge of little or no supervision from agencies thus breaching the integrity of the data collected.
Lastly the issue of security of data collected both in the cloud and on ground. To What extent is there access globally. Hacking into cloud for the what reasons as is seen with NASA, Google, wiki leaks and the l
I think CT data sharing is a very splendid initiative especially in Africa where it is common practice that after publishing the CT report, the data are archived and are never used again. This means that there are fewer secondary analyses using CT datasets from Sub Saharan Africa. Many large multi-country, multi-site studies have been conducted in Africa but we do not see such datasets being utilised after the CT report has been published. One area that this can help is the use of drug safety data that are collected during those studies which would be good in improving pharmacovigilance activities in Africa. For instance, there are few drug safety reports from Africa in the international pharmacovigilance databases (WHO/Uppsala monitoring centre) simply because it depends largely on only spontaneous reports (passive reporting). If safety data from later phase CTs and studies (Phase IIIb/IV) are shared and analysed, it would be a big boost to key pharmacovigilance activities like signal generation for rare AEs.
Research is about digging out data and compile it into information so that others could learn from such findings.
Without sharing data, then it defeats the primary objective of
research whether academic or applied it is either to build to the body of knowledge or share new ideas and solve certain challenging issues.
Research is for public good and it should be shared, but research
costs money as such researchers should be rewarded accordingly so that they could put more effort in doing their work.
Thank you Rafaella for your comment - the PLOSMed article is available here:
Data Sharing in a Humanitarian Organization: The Experience of Médecins Sans Frontières
I enjoyed reading the report. However, I think it's worth remembering that there are already excellent data sharing initiatives amongst those working in less well resourced settings outside the US and EU. Those of us running research in these settings endeavour to run these studies well and often have to employ different approaches because we work in low-resource settings and with vulnerable populations. We are already quite good at sharing because rare and neglected diseases necessitates collaboration. To share effectively, however, we need to get better at standardising the data collected or it is not much good for pooled analysis. Furthermore, any proposed solution needs to support research and not create expensive and burdensome barriers. I think most people would support the concept of data sharing, but it has to be made as easy for an independent academic investigator as it is for a pharma company.
Also the comment that 'most clinical trials are global in nature' may be true of trials originating in the US or EU, but is much less true of trials in resource-poor settings. A director of a large research organisation in Asia I can assure the committee that there is plenty of high quality, important work initiated and led within these regions. The IOM's report does not appear to acknowledge this fact; it seems to have been written very much from the perspective of US / EU sponsors.
Clear and standardized ways of collecting and using data need to be developed from the start of every trial, a process which often suffer limitations regarding all components which may be of importance to future research or use. The WWARN has developed a tool for data pooling to address different questions regarding trials of some medication but there need to be more understanding of the ethical and legal dimension involved in sharing trial data. In addition,increasingly trialists are considering embedding qualitative studies within trial designs and this offers an opportunity to stretch beyond the current considerations of quantitative data sharing and standardisation.
All, I agree with the comments of Jim Todd that highlights the risk of low-income setting trials becoming a raw resource with value addition from further analysis taking place in more developed settings. As with many commodities from low-income settings such resources are typically exported for processing elsewhere.
In addition there is one broad risk related to benefit sharing or fairness that has not been mentioned as far as I can see. This is related to the fact that we tend to ignore that we operate within a relatively bounded space when it comes to resources for research - although the total available has improved it is not infinite. Inevitably therefore we have to consider the opportunity costs of 'new cost centres'.
So allocating resources to support data sharing and secondary analysis risks reducing funds available for primary research. This may not be so bad if the secondary research provides answers that are better or quicker than further primary research, perhaps especially for areas like safety of new products. However, there remains an imbalance in what research funds are spent on. I am not sure the numbers have been recalculated but I suspect the 10/90 gap remains; where 90% of the world's health problems receive 10% of the research funding. Funding for data sharing and shared analysis is likely to focus on those areas with all the research - in other words it too will focus on the needs of the 10% and potentially maintain or worsen the inequity in research funding allocation globally.
Linked to the ideas above I think most people appreciate that there is a danger from increased costs of additional regulation in that they make the primary research more expensive and may result in fewer trials being done especially in places that already have limited funding.
I would prefer to see a parallel focus on how to make conducting trials (in the broadest sense not just individual RCT) easier and cheaper and ones that focus on pragmatic designs in really representative settings. As someone who has worked on trying to translate research evidence into policy there is a considerable barrier to uptake of findings because of difficulties generalising findings to real life situations. Data sharing and re-analysis provide a very partial solution to this problem if the trials do not contain appropriate primary data.
I think the report risks overstating the potential harms of data sharing (e.g., pg 13). For example, when we view flawed secondary analyses as a risk, this presumes that the primary analysis is not flawed. Absent transparency initiatives, it's more difficult to have confidence in any analysis.
Also, it's worth mentioning ideas for mitigating other potential harms, such as loss of IP and attenuated motives for private sector innovation. These could be mitigated with exclusivity or emarbgo periods.
Finally, I am skeptical that the financial costs of data sharing will be large compared to the cost of running a clinical trial.
While data sharing is beneficial to the research community at large there is need for the researcher who collected the data to get direct benefit for reuse of this data,especially when new knowledge that would have impact on health care is generated.This would serve as an incentive for sharing of data.
The report touches on data ownership but there is no clear definition of who owns the data and this has a huge impact on effectiveness of guidelines and policies for data sharing.Going by the premise that the data is owned by the study participant, it is critical to ensure that the rights of the participants are protected. Consent should be explicitly sought from participants for sharing of their data while for retrospective data a framework to ensure ethical sharing of data and information dissemination back to the participants should be created.
Please note that we have added two documents to the blog post above, both from Francois Houyez, Director of Treatment Information and Access at Eurordis. These documents are at the bottom of the post.
Thanks for all the comments thus far, which are very helpful. Please do not forget to share with colleagues who have yet to comment.
As someone who has spent a lot of hours producing secondary datasets of low-resource setting trial data for others' projects I would not underestimate the amount of time that this takes, creating meta-data for different specialised analyses that tend to get requested from large trial datasets, responding to queries etc - and then dealing with the governance around making sure reports/manuscripts get sent in with appropriate deadlines for approval, get forwarded to appropriate trial committees/funders etc, comments get sent back, responded to.
These are all administrative tasks, but without them it is difficult to see how the original researchers can be appropriate data custodians on behalf of original trial participants.
The tension of course is that mostly the original researchers would prefer to be doing these secondary analyses themselves, rather than preparing data for others to do this.
What isn't well thought through IMO with regard to non-commercial trials is the fact that most researchers working on trials are on fixed term contracts, with as little resource as possible to keep trial costs down because finances are always a stretch. When these people (Trial Manager, Statistician) leave at the end of the trial, is the next trial supposed to pick up the cost of supporting the previous trial? It's naive to think that there are not ongoing costs associated with data sharing (as above), but I don't see funders committing to long-term support to this. They seem to think that one can just post a simple dataset and that will be enough - in reality, the most interesting questions always arise from the much more complicated secondary data collected within a trial.
I'd like to add my voice to the points raised by Jim Todd and Sarah Walker about the possible imbalance of those with the time and resources to benefit from more open data and those who collect primary data in resource poor settings. I am firmly of the opinion that more widespread use of data has to be good for science but as both Sarah and Jim quite rightly highlight who is going to pay for this so that it becomes a routine part of the work of studies/trials? Furthermore, the people with the real skill to exercise good data management practices within resource poor locations are few on the ground and many alternative career paths exist that are not just more rewarding in a remunerative sense but also have more room for further professional development.
Sharing data is very important now a days to promote clinial trial in low setting resources but other wise quality of data and stanarized data is important. So we have to put in our mind several issues for this training of scientist in low resources countries in data managements,standardized procedure for data setting , and try to compermise the research ethical issue and clarify research agreement and material transfer agreement incourge the funding of data mangement projects or data sharing and biobanking this it can solve the problem of payment and incentives
It is true that it is costly and challenging to collect clinical data in developing countries so it is important that these researchers have a way to monetize sharing of their data to fund more research, and that they are not always dependent on grants. We could imagine a platform to facilitate this data sharing. Not for profit entities could have a way to access some of this data at no cost, while commercial companies would pay a subscription for this access. The revenues will be split between the initial researchers who produced the data and the vendor that would have built this data platform sharing. Adherence to clinical standards such as CDISC is critical to make sense of the data. Data sharing should also extend to CRFs (imagine a case report library) and reports.
This is a very welcome initiative with both potentially beneficial gains and risk concerns!
If well regulated and co-funded by sponsors, this initiative can facilitate greatly the different Networks of Excellence, such as those funded by EDCTP in Africa, to support efficient data management, multi-level analysis and adequately powered evidence for multi-centre clinical trials.
It also facilitates expedited secondary data analysis of multiple databases across trial sites, countries, continents to generate adequately powered evidence for action and knowledge management.
There will be increased potential for capacity-building in secondary data analysis skills and utilization of trial data, especially critical in resource-poor settings.
There are caveats related to: intensive funds and resources involved; donor-indifference to funding such initiatives, inadequate regulatory framework against misuse, undue data mining with high risk of irrelevant/plausible 'significant' p-values and confidence intervals (spurious!) and the annoying non acknowledgement of the contributors of the raw data, especially in resource-poor settings etc.
How shall we guarantee and successfully minimized cyber-security risks involved? What will be the arrangements and potential sanctions/penalties for such abuse?
It is very important that IOM contributes to firm recommendations on the robust guidelines for data sharing; resilient regulatory framework; safe-guards/checks against potential breach of participants' confidentiality, intellectual property rights, cyber-security concerns, plagiarism, spurious/misleading analyses from large meta-analysis.
IOM should also contribute to higher level advocacy for co-funding these initiatives by interested sponsors. Profit making agencies must be charged to access and use raw data from the different sources.
I will share the current IOM draft report within our research/innovation networks and encourage their participation and contributions.
All best wishes!
I'd like to draw your attention to one of the most important issues in data sharing - good data management. In academia, there is a shortage of data management professionals and resources. Good DM principles is key to obtaining good quality data. We are putting the cart before the horse. There is an urgent need to improve data quality before we can discuss the concept of open data.