Clinical Impact Beyond 'Time to First' Analyses

Posted by Esha Senchaudhuri

Oct 21, 2014 9:00:00 AM

Every year, the East Users Group Meeting brings together notable experts from industry and academia to discuss the future of biostatistical advances in clinical trials, as well as the role of software in facilitating these breakthroughs. In honor of this year’s event, which will be taking place at the Loews Hotel in Philadelphia on October 22, the Cytel Blog will spend the next couple of weeks providing glimpses into the range of discussion presented at the EUGM table.

One debate that has already received quite some attention, involves the weighting of various morbidities in studies with composite endpoints. In a 2013 editorial in the European Heart Journal of the European Society of Cardiology, EUGM speaker Professor L.J. Wei and his colleagues wrote, "A reported P-value must not be confused with an assessment of the magnitude of a treatment's effect in a way that is meaningful to the patient, the clinician and the regulator." [1]

As a toy example, imagine that you are a physician who has to prescribe one of two (hypothetical) drugs to an ailing patient. The first drug, call it Drug A, reduces pain moderately and extends a patient’s life by six years. The second drug, Drug B, eradicates pain for two years, but generates a 50% chance of stroke in the third year, and has never been known to extend a patient’s life by more than five years. Even in this artificially crafted example, asking which treatment is preferrable would require a careful weighting of QALYs and DALYs. Even then, it may not generate a calculation that is helpful for a patient who is considering his options. 

In the 2013 editorial, Professor Wei and colleagues identify two difficulties with the method of attaching severity-weights to various morbidities, as proposed by Bakal et al, in 2012 [2]. The first is the fact that adding severity-weights to non-fatal endpoints may still fail to capture all clinically meaningful information. Consider again the stroke suffered by approximately 50% of patients who take Drug B. If we take the stroke to be a non-fatal endpoint in a study of the drug, should we not distinguish between someone who suffers a stroke and never recovers, and another who suffers a stroke and makes a full-recovery? If so, the authors conclude that, “[W]e will need to leave our current statistical/regulatory comfort zone by acknowledging that the current paradigm of ‘time to first’ analyses ignores a great deal of clinically relevant information.” [1]

However, if we want to make this distinction, what other distinctions do we need to consider in a full analysis of the benefits and costs associated with a drug like Drug B? The authors argue that at least all adverse events which can be prevented should be considered for a risk-benefit analysis.  

This in turn raises a second challenge, since in many cases there might be a trade-off between completeness of information, and what may be called ‘clinical interpretability,’ where clinical interpretability signifies the degree to which statistical findings can be understood in a way important for a clinician and patient. For example, if we take mortality to be the only endpoint that we study, the results are easy to interpret, but they fail to provide all relevant information (e.g. information about morbidities.) On the other hand, if we provide a detailed account of every single adverse event suffered by a patient, we may not be able to come up with a sensible statistical summary which can interpret this range of data in a meaningful way.

Despite these challenges, Wei and his co-authors point out that several novel approaches in clinical biostatistics aim to provide better quality information about clinical impact. They cite twelve modern approaches, all of which accept the elemental necessity of moving 'beyond the comfort zone' of traditional clinical outcome trials. 

We look forward to hearing more about what these approaches are and how clinical trial design can help overcome this roadblock, at the East Users Group Meeting on Wednesday.  


Related Items of Interest

[1] Brian Claggett, LJ Wei and Marc A. Pfeffer, 'Moving Beyond our Comfort Zone,' European Heart Journal 2013.

[2] Jeffrey Bakal, et al., 'Evaluation of early percutaneous coronary intervention vs. standard therapy after fibrinolysis for ST-segment elevation myocardial infarction: contribution of weighting the composite endpoint,' European Heart Journal 2012. 

East Users Group Meeting 2014 Agenda

 East-Banner-6.3

Topics: East, East 6.3, Statistical Programming, Statistical Puzzles, Statistical Analysis, Cardiovascular

The Cytel blog keeps you up to speed with the latest developments in biostatistics and clinical biometrics.  Sign up for updates direct to your inbox. You can unsubscribe at any time.

 

Posts by Topic

see all

Recent Posts