TL9: Benefit-Risk Assessment via Responders in the Absence of Established Definition
*Amarjot Kaur, Merck Research labs 

Keywords: benefit-risk, number needed to treat, number needed to harm, responders

"Discussion Leaders: Amarjot Kaur (Merck Research Labs) & Tammy Massie (CBER, FDA)

Abstract: There are many qualitative and quantitative approaches available for making the benefit-risk assessment of a given investigative treatment or across different treatments. Number needed to treat (NNT) and number needed to harm (NNH) is one of the quantitative approach for the relative assessment of efficacy and safety of a given treatment. The NNT represents the number of subjects who need to receive a treatment for one additional person to benefit from it, while the NNH represents the number of persons who must receive the treatment for one additional person to experience harm. The NNT calculations require proportion of responders based on a priori definition of some dichotomized response which may or may not be readily available depending upon the type of response and the therapeutic area. For example, the definition of responders is not straight forward for continuous efficacy variables in situations when there is no well-established minimal clinically meaningful response and/or in situations where the baseline is absent. The NNH calculations require the proportion of subjects with safety events of interest which are typically available as a dichotomized response. In the absence of an established definition of responders for NNT analysis one could consider alternative reasonable approaches such as:

1) Consider variety of clinically meaningful cutoffs to define responders and examine the corresponding variation in NNTs as a way to examine the sensitivity around proposed definitions.

2) Consider a reasonable estimate of clinically important response as a proportion of the endpoint range to define responders, e.g., 10% of the range of the endpoint assuming a linear response(portnoy et al. 2004).

In this roundtable discussion, we would like to explore and share ideas of defining responders for the efficacy variables in situations when there is no well-established definition (e.g. patient reported symptom scores in the allergy immunotherapy trials). We believe this discussion will be very helpful in providing future guidance on this topic. "