I would add the proviso that IRB approval is necessary. Once there are standard, multiple agents proven by RCT of an active drug vs placebo for a given condition, then IRB may not consider a RCT to be ethical if the condition you are treating should not have a placebo group (depends upon condition, availability of other agent s already proven efficacy in RCT.) They will allow comparison between proven active agent and novel agent in some cases. This can be challenging to recruit especially if patients are naive to exposure to prior known agents, Patients may be willing to get randomized with full Informed Consent . If prior exposure, especially if ADR or AE occurred they may or may not be willing to risk randomization (depending upon the nature and severity of the AE). That group , failure to known agent (LOE) or AEs, may be ideal to consider entering into a novel RCT with active agent vs placebo if there is no alternative approved treatment (in another class that is not contraindicated depending on the prior ADR in a particular of patient ).

That is why I never say "always" "without exception".

Food for thought.

RCTs are the golden standard for studies comparing interventions. Exceptions could be done taking the risk that the study will be considered not acceptable by regulatory authorities or the scientific community. Usually this risk can be accepted for preliminary small scale studies (even if non randomization will remain a controversial fault for your trial) provided that the IRB approve the study protocol. Despite randomization other factors should be considered. What about the study design? And what about masking procedures? I think that the answer to this question is complicated by different aspects related to the intervention and the study population. On the other side if the randomization procedure is not implemented for a test and treat study you could face with the same problems if you carried out a non randomized study (groups unbalancement is always behind your door).

interesting and valuable question. My expertise is in wound care and there is now an emerging field of wound diagnostics. The issue is: does testing and then treating based on the diagnostic improve healing outcomes? there is little if anything done on this, yet such devices are growing in popularity and costs. I believe the RCT is the best answer with which to address these questions that you have posed, but it should also consider the issue of surrogate endpoints and how they relate to the 'test' option. By aiming to use the RCT one is aiming for the highest level of evidence and although challenging in attempting to answer your question are not impossible. Good luck

This is an important question, as most test and treat studies originate within the clinical application setting. My field is pre-hospital emergency care, an area which highlights the challenges of RCT studies. Firstly, the pre-hospital field is not well funded, and attracting funding is challenging in order to perform RCT's. The nature of pre-hospital practice also inherrently involves issues with ethics approval (specifically the nature of informed consent) for studies. And whilst the puritans see the RCT as a gold standard, there is validity in accepting lesser quality observational studies in a practical setting in order to provide a level of evidence where perhaps there is none, particularly as this type of research is often applied practically and within a quality assurance aprroach to clinical practice. I too believe that the RCT is the gold standard of study design, and yet where circumstances preclude the funding, conduct or ethical nature of the research, pragmatism would suggest that we should not be too hasty in dismissing the lesser forms of research (particularly if they generate interest that leads to an RCT). I am reminded of this too well every time I see the word "paucity" attached to pre-hospital research.

Test and treat, as you say are probably best investigated in an RCT design. However, if you are testing a general treatment approach, then it would not matter how individual test results could affect the choice of exact treatment choices within the general appraoch. This would probably be true quite often, as in the real clinical world, many clinicians do not simply administer one specific treatment but the treatment is encompassed in a care giving program.

Another consideration that might help you, in addition to the other relevant comments, is to consider the test and treat interventions as strategies, allowing for some flexibility in the actual interventions being carried out. All patients start with the test, and conditional on the outcome, the preferred treatment is indicated, but in more or less defined situations, a deviating treatment may be given (being an alternative treatment, the treatment in the other arm, etc).

Such a pragmatic perspective will make it easier to evaluate your question with an RCT , as both patients and doctors will be more likely to participate and/or comply with the protocol. Also the results will better reflect actual practice and thus the actual outcomes that can be anticipated, and once the evidence is there, it is often easier to implement the superior strategy in clinical practice.

RCT is envisageable under the condition of equipoise.[1] In other words, there is no existing evidence that the treatment is better than its absence, only suspicions.

This means that when it is evident positive effects overweigh negative effects, RCT are not necessary. There are two nice exemples of such situations:

1. Parachutes preventing injuries from high falls.[2]

2. Relying on glasses to correct for refractive errors.

References

1. van der Graaf R, van Delden JJ. Equipoise should be amended, not abandoned. Clin Trials. 2011 Aug;8(4):408-16. doi: 10.1177/1740774511409600. PMID: 21746767

2. Smith GC, Pell JP. Parachute use to prevent death and major trauma related to gravitational challenge: systematic review of randomised controlled trials. BMJ. 2003 Dec 20;327(7429):1459-61. PMID: 14684649

I would add three points:

1) RCTs have taken a key role in the evaluation of screening programmes, which would classify as test-treat interventions as you describe. The lessons learned in the RCTs of mammography (where there is current controversy of its benefit), should make it clear that RCTs are definitely required, and that they need to be done and reported to a very high standard.

2) Tests improve outcomes in several different ways. If the outcome is improved because a new test is more accurate than the test it replaces, the resulting impact on patient outcomes will be very small, as only those who would get discrepant test results on the two tests will get different patient management as a result, and only a proportion of those would benefit from treatment. The bias in non-randomised studies may dominate any observed treatment effect, supporting the need for RCTs (but they probably would need to be larger than those typically done).

3) We have reviewed >100 test-treat RCTs as part of a project, and will shortly be reporting our findings concerning their design and delivery – we have observed that few give adequate descriptions of the “protocol” followed in terms of the testing and associated interventions in each arm to understand what the test-treat intervention is that is being evaluated. This may be because it cannot be protocolised, but it is a major barrier to the future application of the study results. If you can’t determine what the intervention is (including how clinicians use test information to form management plans), then there is little point in undertaking an expensive and difficult study.

Useful link to our research BMJ framework paper on these issues http://www.bmj.com/content/344/bmj.e686

Hi,

I understand that RCTs are the gold standard but RCTs are not the answer to all problems. Long term outcomes are best observed under observational settings. I am working on many commercialisation strategies where we have used data from nRCTs and observational studies to find the comparative effectiveness. In some cases even it is not feasible to conduct a RCT.

For further details around comprative effectiveness research please contact on [email protected].

Best wishes,

Kashif