Desktop version

Home arrow Mathematics

  • Increase font
  • Decrease font

<<   CONTENTS   >>


Table of Contents:

An important activity for both the pharmaceutical company and the regulators is product promotion and medical communication. This activity is highly regulated and highly scrutinized. If not done properly there can be major consequences to public health and to the reputation of the company, as well as financial penalties. On the otherhand, clinical trials contain much information outside of the primary endpoint and it is helpful to prescribers, payers, and patients to communicate this information. This point of view has been recognized by FDA in their 2018 guidance, “Medical product communications that are consistent with the FDA-required labeling.” This guidance reduces the degree of support for a promotional piece from “substantial evidence” to “sufficient evidence” that is “scientifically appropriate and statistically sound.” However, the material will be judged as not acceptable if it leaves a misleading impression. The need for statistical input in this area is self-evident. The following are general statistical points that should be considered on review of promotional material in order to assess whether the material could be considered misleading.

  • • The treatment effect or information that is promoted should be derived from all studies that address the research question, i.e., it is not seen in a single study if more than one study addresses the issue -inclusive, not selective of positive findings. Of course, there may only be one study that addresses the issue in which case it should be statistically strong.
  • • The effect among the studies should be consistent, i.e., heterogeneity of effect is small. One should consider whether the material is consistent with the totality of the sponsors data.
  • • To address the potential for the criticism of post hoc analyses, in the case of multiple secondary endpoints, there should be an appropriate multiplicity procedure to control the overall Type I error, which is a major regulatory concern.
  • • If there is no multiplicity issue for a particular finding, then the clinical importance of the finding and any prior justification or information to support the finding would support the promotional material. In this regard, the results are clinically important (informative to prescriber, patient, payer) and hence statistically stronger (viewed as less post hoc) if the analyses address, for example, aspects of the primary endpoint, components of a composite primary endpoint, dependence of response on disease severity, onset/duration of effect, or the treatment effect in clinically important subgroups. Although FDA has frowned on the promotion of subgroup effects in the past, the current guidance specifically allows for subgroup results under certain circumstances. This is consistent with the development of targeted therapy, for example, in oncology, where the results in tumors with or without a genetic marker would be important.
  • • In reviewing promotional material, the statistician should be cognizant of the clinical context and prior justification for the material, so it is not criticized as data mining.

In the guidance, for promotional material to be not misleading, FDA stresses that product communications should not overstate the findings or the conclusions that can be drawn from such studies or analyses. The presentation requires proper contextual language including limitations of the strength of evidence (FDA 2018). This is where statistical input is critical to get the context correct.


The main purpose of this chapter is to describe behaviors that can lead to a strong strategic role for the statistician internally and in regulatory and other external interactions. The statistician should establish a strategic and influential role internally so that there is a natural inclusion of the statistician as a strategic player in regulatory interaction. While the recommendations given in this chapter are intended to have general application, the statistician should interpret them in the context of the specific regulatory setting. As indicated in the introduction, the importance of statistics and statisticians has been firmly established in the pharmaceutical industry. It will benefit the drug-development process to extend that importance into the regulatory sphere.


DeMets, D., Furberg, C., Friedman, L. (2006) Data Monitoring in Clinical Trials: A Case Studies Approach. New York: Springer.

Ellenberg, S., Fleming, T., DeMets, D. (2002) Data Monitoring Committees in Clinical Trials: A Practical Perspective. Hoboken, New Jersey: John Wiley & Sons.

Emir, B., Amaratunga, D., Beltangady, M. et al. (2013) Generating productive dialogue between consulting statisticians and their clients in the pharmaceutical and medical research settings. Open Access. Medical Statistics . 3: 51-56.

European Medicines Agency (2005) Guideline on data monitoring committees,

Food and Drug Administration (2006) Guidance for clinical trial sponsors. Establishment and operation of clinical trial data monitoring committees.

Food and Drug Administration (2018) Guidance documents/Medical product communications are consistent FDA-required labeling questions and answers.

Grieve, A. P. (2002) Do statisticians count? A personal view. Pharmaceutical Statistics. 1(1): 35-43.

Grieve, A. P. (2005) The professionalization of the “shoe clerk”. Journal of the Royal Statistical Society Series A (Statistics in Society). 168(4): 639-656.

Herson, J. (2009) Data and Safety Monitoring Committees in Clinical Trials. New York: Chapman 8c Hall/CRC.

Lewi, P. J. (2005) The role of statistics in the success of a pharmaceutical research laboratory: A historical case description. Journal of Chemometrics. 19: 282-287.

Marquardt, D. W. (1987) The importance of statisticians. Journal of American Statistical Association. 82(397): 1-7.

Rockhold, F. W. (2009) Strategic use of statistical thinking in drug development. Statistics in Medicine. 19: 3211-3217.

Unwin A. (2007) Statistical consulting interactions: A personal view. Advances in Statistical Analysis. 91(4): 349-359.

<<   CONTENTS   >>

Related topics