We are writing a certain protocol for use in a clinical research setting. During literature search, the committee wondered one of the trialists proposing for a RETROSPECTIVE design in the proposed trial with a very clear view that it will save costs!
Since that sounded like a total confusion, no one wanted even to discuss that issue but when given the space to air out his views, the committee member explained it as a possibility and when cornered to deliver evidence of such a wonderful concept he could provide one which even got published in a medical literature which is peer reviewed.
Honestly, from logic of TRIAL design (especially methodology!) how can one analyse a randomised controlled trial with a retrospective design?
Kindly offer your experience regarding this otherwise very appealing issue that might have gone out my mind back then in clinical trial sessions!
Dear Martin,
kindly have a check on this article (available on this link : http://www.ncbi.nlm.nih.gov/pubmed/19032126)
I am still confused (from a point of view of the methodology of trials!) but would wish to know the possibilities for this to be VALID.
Dear Colleagues
By definition a randomised controlled trial is prospective.
Other variants would come under 'pseudo randomisation'.
Indeed many respected journals would expect a trial registration number 'a priori' before participants are enrolled into the experiment.
There are dozens of different definitions on “prospective vs retrospective”, these terms can therefore better be avoided or should be accompanied by the definition used.
Randomisation is used in an effort to prevent an association between prognosis and assignment of treatment. In some specific situations, like genetic variability, it is possible to use previous collected clinical data and check these with newly acquired genetic information and analyse this as a (pseudo) randomized trial.
There is one such design. You give treatment to everyone and measure the outcome. You then do their genetics and test the hypothesis that the gene of interest affects the outcome. This is Mendelian randomisation. There are several examples in the literature. Some have done this in observational studies where the genetic information is subsequently added. Thus the study could be called randomised but retrospective but in practice Mendelian randomisation is a better term.
If the interest is in saving costs over the traditional RCT, observational studies using instrumental variables (IV) or propensity score matching may be helpful.
Examples:
Propensity score matching: http://www.ncbi.nlm.nih.gov/pubmed/21199971
IV: http://www.ncbi.nlm.nih.gov/pubmed/22778301
When you look at the definition of a clinical trial by World Health Organization, it is stated that:
For the purposes of registration, 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.
Thus, by definition a clinical trial is prospective.
Dear Kelvin:
There are 8 basic types of studies, according to 4 main points of view:
1) Experimental vs. observational
2) Longitudinal vs. cross-sectional
3) Prospective vs. retrospective
4) Descriptive vs. analytical (comparatives)
By definition, a randomized clinical trial (RCT) is an experimental + longitudinal + prospective + analytical study.
Experimental because there is a control over exposures (treatments) and randomization is to ensure that the prognostic factors (known and unknown) are distributed equally in the groups.
Longitudinal because the main variable is measured as "before and after" and you perform a follow-up.
Prospective because you start data collection before both exposure and outcomes occur.
Analytical because you compare two or more groups.
It is not possible to perform a retrospective RCT.
You can design a retrospective cohort study to estimate a benefit but not to determine efficacy of two treatments.
Cohort studies (prospective or retrospective) are observational studies.
Regards,
JC Maldonado
As a student of Health Research Methods, I know that when we refer to an RCT, we mean a prospective design which is the reason I do not encourage use of the term "prospective" with RCT because it is already implied.
Actually yes,
Please take a look at this article:
JAMA Oncology. February 23, 2017
Association Between Telomere Length and Risk of Cancer and Non-Neoplastic Diseases. A Mendelian Randomization Study
Actually I am confused, it is known that there is a controlled trial with a historical control from hospital records for example. So , what about having two different interventions for camparing their outcomes taken from hospital records with all needed information taken retrospectively. This design actually I read it , comparing two different techniques for occluding patent ductus A. in pediatric patients. Some papers wrote this design as a retrospective cohort , others wrote it as a retrospective comparative study. I think against my beleifs, it should be considered a clinical trial taken with two historical intervention groups retrospectively.
Since randomization would no be possible in a retrospective study, RCT could not be retrospective. For non-randomized trials, a prospective monitoring of an intervention should be applied and the benefits of clinical trials comes from this characteristic. Otherwise, it couldn't be considered as an interventional study. In other words, when you retrospectively compare two groups who took two different medication, you don't have any authority on the intervention. You are infact an observer of what is being running. Among observational studies, it resembles to a historical cohort rather than a case-control study because the key element of the grouping is not the outcome, but is the influencing factor. However, it could also be considered as a cross-sectional observation for a short course study. All in all, it is not a preferred research method and I don't encourage you to take this method due to the potential sources of bias.
Randomized clinical trail is a prospective study as how can you randomize the study subjects retrospectively?
Since 2005 International Committee of Medical Journal Editors (ICMJE) has required registration of clinical trials before the first participant enters the trial (prospectively). Until now there are some clinical trials published, which are registered later on (retrospectively) or not at all (Harriman and Patel 2016, Article When are clinical trials registered? An analysis of prospect...
).As retrospective registration introduces potential bias, there is ongoing editor's dilemma whether to consider or not a manuscript with such retrospective registration for publication ( https://blogs.biomedcentral.com/bmcblog/2016/04/15/retrospectively-registered-trials-editors-dilemma/ )
I cannot see how RCT's can be retrospective due to the randomisation aspect.
The other thing is reanalysis and reuse of retained biological samples collected during clinical trials. See e.g.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7026937/
- Badges
- Science topic
- Similar topics
- Thesis Research