Remember to register your clinical trial at the start: A note to the SSAI membership from the Acta Anaesth Scand Editor in Chief
Hello fellow SSAI members, readers and contributors to our journal, Acta Anaesthesiologica Scandinavica!
In the salutation, I write that this is our journal. It is so. The SSAI owns the journal, and the SSAI is a membership organization. As SSAI members, we collectively ‘own’ the journal. As consumers of medical research, we have noted the discussion about why many trial results are difficult to reproduce, and if they are reliable. This note focuses on one aspect of improving clinical trial reporting.
First, clinical trials are important for us as clinicians. When there is a clinical question that we want to test, there is no better way to do this than by conducting a clinical study where we prospectively compare the treatment or intervention in question (done in a controlled fashion) to the appropriate comparison treatment. Or, maybe we just test the treatment with no comparison. Ideally, to eliminate as much as possible confounding factors for which we cannot control, we randomly allocate the research subjects (or randomize) to a test treatment or a comparison treatment. That subjects go by chance to one treatment or the other is a powerful way to reduce the impact of factors related to subject inclusion which might disturb the results. If possible, we further try to reduce potential influence of confounding factors by keeping both subject and (if possible) the treating medical teams in the dark (blinding) about which treatment the research subject is receiving (every subject gets as much as possible the same attention). We call this type of effort a clinical trial, and often include the obvious qualifiers when appropriate in describing the effort, such as prospective, randomized, blinded, and more. And, more each year, these types of studies are trying to be ‘pragmatic’, or adapted to the realities of the clinical workplace. We should all try to contribute to these kinds of efforts as much as we can, to understand how we can better treat, and also prevent unnecessary harm, for our patients. At the least, all of us are also consumers and translators/implementers of treatments addressed by these kinds of research findings.
This note today is intended as another attempt to spread the word about a subject around which there has been a bit of confusion for some. When does one need to register their study in a clinical trial registry, and why does this need to be done prospectively (before any subjects are enrolled)? This was not ‘required’ 15 years ago. This may seem like extra work, but it is now a requirement for publishing clinical trials by most highly respected clinical journals.
First, some definitions. There can be some confusion about what constitutes a clinical trial in the first place. At least for purposes of registration, the International Committee of Medical Journal Editors (ICMJE) informed the World Health Organization definition of a clinical trial:
‘…a clinical trial is any research study that prospectively assigns human participants or groups of humans to one or more health-related interventions to evaluate the effects on health outcomes. Clinical trials may also be referred to as interventional trials. Interventions include but are not restricted to drugs, cells and other biological products, surgical procedures, radiologic procedures, devices, behavioural treatments, process-of-care changes, preventive care, etc. This definition includes Phase I to Phase IV trials.’
(http://www.who.int/ictrp/en/)
The key here is that there is a prospectively controlled intervention (allocation to a controlled treatment). Authors do not always describe their study in their manuscript as a clinical trial, even when a journal may recognize the clear hallmarks of a prospective clinical trial. There needs to be clarity about this among all clinical researchers.
Then, which trial registries are correct and should be used for trial registration? A trial registry is a place where trial details are ‘published’ or made available, and approved registries are set up based on an internationally agreed-upon set of information or design. For Nordic researchers, there may now be some automatic trial registration when clinical studies involve drugs or medical devices (see below). I have overhead some complain that some of the very established trial registries require lots of work to enter study details (read: they are detailed!). But, we should not shy away from a bit of work, without which our study may never be publishable in our favourite journal.
Our journal started requiring prospective clinical trial registration for trials that started enrolling subjects since the start of 2010, and Lars Rasmussen as Editor in Chief at that time was instrumental in modernizing our journal and editorial process in this way. Even though sometimes recognition and understanding of this kind of requirement takes time to disseminate throughout the clinical research world, it was not new in 2010. The International Committee of Medical Journal Editors (ICMJE) published this recommendation in 2004 (JAMA 2004; 292: 1364). In this recommendation, there is also a specification for the essential and necessary information to be included in this trial registration (the intention is to include vital aspects of the trial design). The reason for this coming about was that clinical research leaders has started realizing that many clinical trials that were being published were heavily influenced by different forms of bias; they were presenting findings that were not reproducible, and steps needed to be taken to improve the quality of clinical trial reporting. This was one important step. Not by any means the only step needed (probably better research training for clinical researchers is the most important aspect), but still an important step.
What good is derived from prospective clinical trial registration? First, research should be conducted openly. There are arguments against this from some researchers, saying that when the plans for a study are openly available, some unscrupulous researchers may steal the idea and go perform and publish the study before the first group is able to do the same. These registrations are dated, and copying or stealing research ideas is not something that an honourable researcher would like to be accused of, so it is unclear if this is an actual problem.
Another reason for registration is that this encourages the research teams to develop their analysis plans as much as possible before starting their data collection. Clinical trials should not be primarily explorative. They should be based on a specific test of a well-defined question. This is, in fact, one of the limitations of simpler clinical trials, that they only may confidently answer only one specific and limited question (at best), and do not answer several closely related questions which are also relevant (the findings and conclusions are often limited or restricted, and may not be widely applicable).
Additionally, when defining how the research team plans to analyse the results, researchers will specify this at the start by defining primary outcomes, based on which comparisons between groups will be the hypothesis testing. Unfortunately, in years past, when the primary outcome was not prospectively specified, this left open an option to ‘shop’ a bit among the different findings, and use ones that showed positive results to test a hypothesis (potentially post-hoc hypothesis-making), while not necessarily reporting that the original primary outcome did not show a positive result for the tested treatment. Unfortunately, many positive findings occur just by chance (not a real treatment effect), and cannot be reproduced. This is the reason that unexpected positive findings which were not original primary outcomes need to be interpreted carefully. When this kind of potential ‘over-interpretation’ has happened, it is now better understood is that these findings and publications have generally demonstrated in part what the authors had for own biases rather than what would be reproducible for the main hypothesis testing when done correctly. So, sticking to the original study design primary outcomes is very important when reporting clinical trials. The prospective trial registration allows a reader or reviewer to confirm that the primary hypothesis and primary outcomes in the manuscript are the same as in the original study plan.
Finally, registering trial protocols prospectively allows some follow-up for trials that do not get published. Negative trials concerning medical products (including drugs) often have not been published, though the findings often are available.
Even despite these kinds of measures, there are always some methodological limitations for any study, and clinical trials can be published that when examined closely by many, reasons may come up that lead to low confidence in the positive effect estimates. The pressure on authors and journals to publish positive findings has been unreasonable and unfortunate, pressure coming from competition for jobs (need publications on CV’s) and for journal competition (bibliometrics like impact factor leading journals to succumb to pressure to publish articles that they think will be highly cited in order to remain commercially competitive). There has been immense pressure to publish positive results (publish them first) even when the likelihood that those results were reproducible were not very high. Also, there has generally been limited interest in publishing negative results or confirmatory results, since these are unlikely to be often cited. Not an optimal situation for open and objective dissemination of science.
How study findings are published is also changing. Just as clinical trial protocols are widely available at the start, detailed findings from studies, and I mean the raw data, are now often made available with publication of a trial. These are practical now to present as supplementary files for digital versions of the publications. Soon, we probably will only be looking at digital versions of publications. The availability of the raw data for a study means that other researchers can repeat analyses of this, or do new analyses on this already collected and presented data. Look for this! Secondary findings (non-hypothesis testing results) often can help in defining new studies in the same field.
The trial registers and study plan registration are free. In Europe and the European Union starting in 2013, studies involving drugs get registered by the country’s own national medicines agency, with that information being automatically stored ultimately in the EudraCT and EU Clinical Trials Registry database. Many countries have their own trial registries, and the idea is that they are open, searchable, and will contain the relevant information that a critical reader or reviewer will want to find.
I hope that this brief note helps clarify some issues for some who were wondering, at least concerning what constitutes a clinical trial and why prospective trial registration is now required. Our journal has a mission: to promote and publish good medical research and knowledge in our field and in our region. The journal strives to be completely consistent and fair in these requirements. We also have a mission to help all researchers in the Nordic countries be informed about relevant issues, and here about the best processes for performing and publishing clinical trials. In our Nordic region, we have unusually good conditions for conducting multi-centre clinical trials, given the extremely high degree of good follow-up for research subjects. As 5 nations, we may not be optimally taking advantage of this desirable circumstance at least as far as international collaboration, and probably could be performing more multi-centre clinical trials than are occurring at present. The SSAI offers research grant money to support this kind of Nordic collaboration!
As always, at your service,
Michael Haney
Umeå, June 2017