Sample Size and Power Calculations for Additive Interactions

Abstract Interactions measured on the additive scale are more relevant than multiplicative interaction for assessing public health importance and also more closely related to notions of mechanistic synergism. Most work on sample size and power calculations for interaction have focused on the multiplicative scale. Here we derive analytic expressions for sample size and power calculations for interactions on the additive scale. We give formulae for detecting additive interaction on the risk scale from a cohort study, formulae for detecting additive interaction using the relative excess risk for interaction from a logistic regression with cohort data, and formulae for detecting additive interaction for the relative excess risk for interaction from a logistic regression with case-control data. When main effects of both exposures are positive, power to detect positive interaction on the additive scale will be greater than that on the multiplicative scale. Excel spreadsheets are provided for power and sample size calculations for additive, multiplicative, and case-only interaction estimates.

[1]  Sample size and power calculations for case-only interaction studies. , 2011, Epidemiology.

[2]  N E Day,et al.  Sample size determination for studies of gene-environment interaction. , 2001, International journal of epidemiology.

[3]  James M. Robins,et al.  Empirical and counterfactual conditions for sufficient cause interactions , 2008 .

[4]  W D Flanders,et al.  Case-only design to measure gene-gene interaction. , 1999, Epidemiology.

[5]  Tyler J VanderWeele,et al.  Statistical Applications in Genetics and Molecular Biology Epistatic Interactions , 2011 .

[6]  Robert W. Mc Collum.  Epidemiologic Methods , 1961, The Yale Journal of Biology and Medicine.

[7]  S Lemeshow,et al.  Confidence interval estimation of interaction. , 1992, Epidemiology.

[8]  T H Beaty,et al.  Minimum sample size estimation to detect gene-environment interaction in case-control designs. , 1994, American journal of epidemiology.

[9]  T. VanderWeele,et al.  Interpretation of Subgroup Analyses in Randomized Trials: Heterogeneity Versus Secondary Interventions , 2011, Annals of Internal Medicine.

[10]  N. Jewell,et al.  Some surprising results about covariate adjustment in logistic regression models , 1991 .

[11]  W D Flanders,et al.  Sample size requirements in case-only designs to detect gene-environment interaction. , 1997, American journal of epidemiology.

[12]  W. Gauderman.  Sample size requirements for association studies of gene-gene interaction. , 2002, American journal of epidemiology.

[13]  Tyler J Vanderweele.  Inference for additive interaction under exposure misclassification. , 2012, Biometrika.

[14]  M. Graffar.  [Modern epidemiology]. , 1971, Bruxelles medical.

[15]  Jack A. Taylor,et al.  Non-hierarchical logistic models and case-only designs for assessing susceptibility in population-based case-control studies. , 1994, Statistics in medicine.

[16]  P S Albert,et al.  Limitations of the case-only design for identifying gene-environment interactions. , 2001, American journal of epidemiology.

[17]  T. VanderWeele.  The identication of synergism in the su¢ cient-component cause framework , 2007 .

[18]  Onyebuchi A Arah,et al.  Bias Formulas for Sensitivity Analysis of Unmeasured Confounding for General Outcomes, Treatments, and Confounders , 2011, Epidemiology.

[19]  Eugene Demidenko,et al.  Sample size and optimal design for logistic regression with binary interaction , 2008, Statistics in medicine.

[20]  Til Stürmer,et al.  Flexible matching strategies to increase power and efficiency to detect and estimate gene-environment interactions in case-control studies. , 2002, American journal of epidemiology.

[21]  Pascal J. Goldschmidt-Clermont,et al.  Sample size to test for interaction between a specific exposure and a second risk factor in a pair-matched case-control study. , 2000, Statistics in medicine.

[22]  S Greenland,et al.  Power, sample size and smallest detectable effect determination for multivariate studies. , 1985, Statistics in medicine.

[23]  T. VanderWeele.  On the Distinction Between Interaction and Effect Modification , 2009, Epidemiology.

[24]  S Greenland,et al.  Concepts of interaction. , 1980, American journal of epidemiology.

[25]  J H Lubin,et al.  On power and sample size for studying features of the relative odds of disease. , 1990, American journal of epidemiology.

[26]  D. Spiegelman,et al.  Power and sample size calculations for case-control studies of gene-environment interactions with a polytomous exposure variable. , 1997, American journal of epidemiology.

[27]  Stephen Gruber,et al.  Accounting for error due to misclassification of exposures in case–control studies of gene–environment interaction , 2008, Statistics in medicine.

[28]  Tyler J. VanderWeele,et al.  Empirical tests for compositional epistasis , 2010, Nature Reviews Genetics.

[29]  Donald B. Rubin,et al.  Formal modes of statistical inference for causal effects , 1990 .

[30]  P. Phillips.  Epistasis — the essential role of gene interactions in the structure and evolution of genetic systems , 2008, Nature Reviews Genetics.

[31]  Tyler J VanderWeele,et al.  Sufficient Cause Interactions and Statistical Interactions , 2009, Epidemiology.

[32]  W James Gauderman,et al.  Sample size requirements for matched case‐control studies of gene–environment interaction , 2002, Statistics in medicine.

[33]  Hongyu Zhao,et al.  Sample size needed to detect gene-gene interactions using association designs. , 2003, American journal of epidemiology.

[34]  S Greenland,et al.  Tests for interaction in epidemiologic studies: a review and a study of power. , 1983, Statistics in medicine.

[35]  H. Cordell.  Detecting gene–gene interactions that underlie human diseases , 2009, Nature Reviews Genetics.

[36]  Bhramar Mukherjee,et al.  Sensitivity analysis for interactions under unmeasured confounding , 2012, Statistics in medicine.