You may have heard that eating citrus fruit helps treat scurvy, a disease that causes bleeding gums, bulging eyes, swollen joints, and if untreated, death. But did you know that the link between citrus and scurvy was proven by what many call the first clinical trial?
In 1747, James Lind, a ship’s surgeon, tested his theory for preventing the effects of scurvy by giving 12 sailors different dietary acids. He subsequently found that only sailors who consumed citric acid in the form of oranges and lemons improved.
This trial looked at a sample cohort (12 sailors) from a broader population (all the sailors) and tested whether exposing part of that cohort to an intervention of interest (citric acid), and the other part of that cohort to a different intervention (other dietary acids), resulted in different outcomes (improved symptoms) for the two parts of the sample cohort.
But was it a randomised controlled trial?
Lind’s experiment illustrates some principles of a randomised controlled trial, but not all. The common important features of a randomised controlled trial include:
Additionally, many randomised controlled trials are structured so that participants and/or researchers do not know which participants are part of the experimental group and which are part of the control group. Studies where participants don’t know if they are receiving the intervention are called ‘single-blind’ studies whereas studies where both researchers and participants are unware are called ‘double-blind’ studies.
Participants may be ‘blinded’ because of potential biases they may have towards the intervention. Similarly, researchers may have biases about the intervention, or want to provide it to certain participants and not others. Having both parties unaware of who is receiving what – the double-blind structure – is superior because it reduces the impact of these biases on the outcome of the trial.
Randomised controlled trials are excellent at answering questions about the effects of an intervention on a population. This table shows some example questions that are appropriate for randomised controlled trials:
| Question Type | Study Example | |
|---|---|---|
| Effectiveness | Is a proposed intervention as effective as the established intervention? | This study looked at the effectiveness of play-based interventions for pre-schoolers with behavioural issues. |
| Side effects | Does a proposed intervention have worse side effects than the established intervention? | This study looked at medication intended to reduce blood loss after vaginal delivery. |
| Cost of treatment | Which intervention for a specific condition results in less cost to a healthcare service? | This study compared two methods of repairing aortic aneurysms. |
| Patient adherence to treatment | Which intervention are patients more likely to positively respond to? | This study looked at whether SMS messages improved adherence rates in patients receiving antiretroviral therapy. |
| Duration of effect of treatment | Which intervention provides longer-lasting benefit? | This study looked at treatments for colorectal cancer and measured survival rates among participants. |
It is important for RCTs to begin with a clear understanding of what is being measured and how measures will be taken, and for these to be adhered to throughout the life of the trial. Without them there is the risk of researchers introducing bias by looking for answers that the study was not set up to provide. It is also important for the eligibility criteria of the population of interest to be clear.
In a RCT trial participants are randomly allocated to control or experimental groups. Well-designed trials feature randomisation that is as free from bias as possible. A strong report of a RCT should also include details of the randomisation method in order to be as transparent as possible.
Blinding is when the trial participants and/or the people administering the intervention do not know whether a participant has been allocated to the experimental or the control group. This is done to avoid the influence of bias, following the principle that if an individual knows which group they are in they will act differently as a result. “Single-blind” trials have only the participants unaware of group allocation while “double-blind” trials have both the participants and administrators unaware of group allocation. Blinding of administrators is important in order to prevent them from acting differently in response to awareness of whether they are administering the exposure of interest or not. This difference in action may influence the participant’s behaviour, or measurement of the impact of the exposure, or other factors. This could obscure the true effect of the intervention.
The trial participants have been investigated thoroughly and found to be adequately representative of the broader population they represent while still being sufficiently similar so as to produce research data that is free from confounding factors. The sample cohort also needs to be large enough so that statistical analysis allows for results to be applicable to the eligible population with a reasonable degree of confidence.
It is difficult for a study to be entirely free of bias. However, researchers can design their study in such a manner to control for potential sources of bias as much as possible. This includes appropriately managing any involvement by commercial partners in the trial in order to prevent their participation having any influence on its execution and conclusions.
In order for the trial participants to be a fair representation of the population of interest (i.e. the larger group that the sample is taken from), recruitment of this sample cohort needs to be robust. This means that recruitment should be conducted with the purpose of obtaining a representative cross-section of the population of interest, and that segments of that population should not be overrepresented if that could cause bias in the results. For example, if the proportion of males in the sample cohort is much higher than in the population of interest the sample is representing, this may have implications for the application of the study’s conclusion to the broader population which (proportionally) has far fewer males than the sample that were actually studied.
While randomisation is an important strength of randomised controlled trials, some forms of randomisation are susceptible to bias. For this reason, randomisation needs to be effectively concealed from both the party administering the intervention and the participant receiving it. Effective concealment means that the administrator of the intervention is unable to learn beforehand or otherwise effect which participants receive which treatments.
A study has greater validity if all participants are accounted for throughout the life of the trial, rather than being ‘lost to follow up’ after the intervention and possibly skewing the data as a result. Reporting of trials should account for all participants initially enrolled in the study.
In solid trials, all participants are analysed in the group that corresponds to how they were intended to be treated, and not necessarily how they were treated. In other words, if a participant in the intervention group ended up not actually receiving the intervention, their data should still be included in the intervention groups’ data and not in the control group. This is known as an intention-to-treat analysis and preserves the important randomisation feature of the trial.
Many trials are conducted with the sponsorship of companies that sell the intervention being studied in the trial. Common examples include pharmaceuticals or medical devices. Research published in 2018 found that studies which had industry sponsorship were more likely to report results favourable to the intervention of interest than studies not sponsored by industry. This suggests the existence of a bias for the studies which received industry sponsorship.
Despite this, it is important to be aware that many trials would not happen at all without industry sponsorship. Furthermore, the existence of industry sponsorship bias does not mean that all commercially-sponsored trials are inherently biased.
Evaluation of papers reporting studies is often referred to as critical appraisal. To enable the critical appraisal process, checklists have been developed to guide the evaluator(s) through the process. Critical appraisal checklists do this by prompting to reader to ask certain questions of the paper at hand.
While there are many critical appraisal checklists, it is best to use checklists designed for the particular type of study the paper is reporting. Sources of critical appraisal checklists for randomised controlled trials include:
Finally, while not technically a critical appraisal checklist, the CONSORT guidelines for reporting randomised trials could be useful to help understand the expectations for appropriate reporting of such a study.
Meldrum, M. (1998). "A calculated risk": the Salk polio vaccine field trials of 1954. BMJ, 317(7167), 1233-1236.
This paper details the successful 1954 trial of a vaccine against polio. This remains one of the largest trials ever conducted.
Palm, U., Schiller, C., Fintescu, Z., Obermeier, M., Keeser, D., Reisinger, E., . & Padberg, F. (2012). Transcranial direct current stimulation in treatment resistant depression: a randomized double-blind, placebo-controlled study. Brain Stimulation, 5(3), 242-251.
This study looks at whether transcranial direct current stimulation (tDCS) is effective therapy for treatment-resistant major depressive episodes. In the study tDCS is compared to a placebo stimulation.
Macdonald, G. Z., Button, D. C., Drinkwater, E. J., & Behm, D. G. (2014). Foam rolling as a recovery tool after an intense bout of physical activity. Medicine and Science in Sports and Exercise, 46(1), 131-142.
This study looked at the usefulness of foam rolling as a recovery method after exercise. It randomly assigned participants into two groups who then both participated in the same exercise program, with one group then receiving 20 minutes of foam rolling exercises at the end of the program.
Begg, C., Cho, M., Eastwood, S., Horton, R., Moher, D., Olkin, I., . . . Stroup, D. F. (1996). Improving the quality of reporting of randomized controlled trials. The CONSORT statement. JAMA, 276(8), 637-639.
Bowers, D., Bewick, B., House, A., & Owens, D. (2013). Understanding clinical papers (3rd ed.). Chichester, West Sussex: Wiley Blackwell.
CONSORT Group. (2010). CONSORT 2010 checklist of information to include when reporting a randomised trial. Retrieved March 14, 2019, from http://www.consort-statement.org/
Greenhalgh, T. (2014). How to read a paper : the basics of evidence-based medicine (5th ed.). Chichester, West Sussex: Wiley.
Harris, M., Jackson, D., & Taylor, G. (2014). Clinical evidence made easy. Oxfordshire, England: Scion Publishing.
Hoffmann, T., Bennett, S., & Del Mar, C. (2017). Evidence-based practice across the health professions (3rd ed.). Chatswood, N.S.W., Australia: Elsevier Australia.
Lundh, A., Lexchin, J., Mintzes, B., Schroll, J. B., & Bero, L. (2018). Industry sponsorship and research outcome: systematic review with meta-analysis. Intensive Care Medicine, 44(10), 1603-1612. doi: 10.1007/s00134-018-5293-7
Tulchinsky, T. H. (2018). Case studies in public health. London, England: Academic Press.