gms | German Medical Science

63. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

02. - 06.09.2018, Osnabrück

Sample size reestimation for clinical trials with longitudinal negative binomial counts including time trends

Meeting Abstract

  • Thomas Asendorf - Institut für medizinische Statistik, Universitätsmedizin Göttingen, Göttingen, Deutschland
  • Robin Henderson - Mathematics, Statistics and Physics, Newcastle University, Newcastle upon Tyne, Great Britain
  • Heinz Schmidli - Novartis Pharma AG, Basel, Schweiz
  • Tim Friede - Universitätsmedizin Göttingen, Göttingen, Deutschland

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie. 63. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Osnabrück, 02.-06.09.2018. Düsseldorf: German Medical Science GMS Publishing House; 2018. DocAbstr. 281

doi: 10.3205/18gmds035, urn:nbn:de:0183-18gmds0353

Veröffentlicht: 27. August 2018

© 2018 Asendorf et al.
Dieser Artikel ist ein Open-Access-Artikel und steht unter den Lizenzbedingungen der Creative Commons Attribution 4.0 License (Namensnennung). Lizenz-Angaben siehe



Introduction: In some diseases such as multiple sclerosis, lesion counts obtained from magnetic resonance imaging are used as markers of disease progression [1]. This leads to longitudinal, and typically overdispersed, count data outcomes. The nature of this data should be accounted for in the analysis, and equally in the planning of sample sizes. Models for such data include a number of nuisance parameters, which can be difficult to specify at the planning stage. Therefore, blinded sample size reestimation procedures are considered allowing for an adjustment of the sample size based on estimates of relevant nuisance parameters at an interim time point. To date, the methods available for sample size reestimation have assumed the mean count to be time-constant within patients [2]. We propose a new modelling approach, that maintains the advantages of established procedures while allowing for treatment-specific time trends in the mean response.

Methods: For maximum-likelihood inference methods based on a gamma-frailty model by Fiocco et al. [3] we present sample size estimation and reestimation procedures. Sample size reestimation is performed at an interim time point on blinded data by using a likelihood from a mixture distribution, thereby maintaining the trials integrity. Treatment-specific time trends can be defined in a flexible manner, allowing for a large variety of possible time trends.

Results: Based on the gamma frailty model from Fiocco et al. [3], sample size estimation and reestimation methods are developed which allow for treatment-specific time trends. A simulation study is conducted to assess the effectiveness of blinded sample size reestimation methods over fixed designs. Sample sizes attained through blinded sample size reestimation procedures are shown to maintain the desired study power without inflating the type I error rate.

Discussion: Existing procedures for sample size estimation and reestimation for longitudinal and overdispersed count data are extended to account for treatment-specific trends. The presented procedures are shown to meet regulatory requirements and their usage, as well as limits, are illustrated.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.


European Medicines Agency (EMEA). Guideline on clinical investigation of medicinal products for the treatment of Multiple Sclerosis. 2015.
Asendorf T, Henderson R, Schmidli H, Friede T. Modelling and sample size reestimation for longitudinal count data with incomplete follow up. Stat Methods Med Res. 2017 Jan:962280217715664. DOI: 10.1177/0962280217715664. Externer Link
Fiocco M, Putter H, Van Houwelingen JC. A new serially correlated gamma-frailty process for longitudinal count data. Biostatistics. 2009;10(2):245-57.