Skip to main content

Table 2 Article classifications

From: Statistical methods for the analysis of adverse event data in randomised controlled trials: a scoping review and taxonomy

Authors Year Taxonomy a Further classification variables Brief summary
  V, HT, E, DMP Prespecified or Emerging (single or multiple outcomes) (Group) Sequential (monitoring) - yes/no
Amit, Heiberger & Lane [25] 2008 V Emerging (single & multiple) No Dot plot for emerging AEs, Kaplan-Meier and hazard function for single AEs and cumulative frequency plots, boxplots and line graphs for continuous outcomes
Chuang-Stein, Le & Chen [26] 2001 V Emerging (single) No Displays two-by-two frequencies graphically for emerging AEs, histograms and delta plots for continuous outcomes
Chuang-Stein & Xia [27] 2013 V Emerging (single & multiple) No Bar charts, Venn diagrams and Forest plots for emerging AEs, risk over time for single AEs and e-Dish plots for continuous outcomes
Karpefors & Weatherall [28] 2018 V Emerging (multiple) No Tendril plot for emerging AEs
Southworth [29] 2008 V Emerging (single) No Scatterplot with regression outputs for continuous outcomes
Trost & Freston [30] 2008 V Emerging (multiple) No Vector plots for continuous outcomes, includes 3 outcomes per plot
Zink, Wolfinger & Mann [31] 2013 V Emerging (multiple) No Volcano plot for emerging AEs
Zink, Marchenko, Sanchez-Kam, Ma & Jiang [14] 2018 V Emerging (multiple) No Heat map for emerging AEs
Bolland & Whitehead [32] 2000 HT Prespecified Yes Alpha spending function
Fleishman & Parker [33] 2012 HT Prespecified Yes Alpha spending function, adjustment to significance threshold and conditional power
Lieu et al. [34] 2007 HT Prespecified Yes Likelihood ratio test
Liu [35] 2007 HT Prespecified No Non-inferiority test
Shih, Lai, Heyse & Chen [36] 2010 HT Prespecified Yes Likelihood ratio test
Agresti & Klingenberg [37] 2005 HT Emerging (overall profile) No Multivariate likelihood ratio tests for overall AE numbers
Bristol & Patel [38] 1990 HT Emerging (overall profile) No Multivariate likelihood ratio test with Markov chains for overall AE numbers, incorporating recurrent events
Chuang-Stein, Mohberg & Musselman [39] 1992 HT Emerging (overall profile) No Multivariate test for overall AE numbers with chi-squared distribution, incorporating severity and participant acceptability scores
Huang, Zalkikar & Tiwari [40] 2014 HT Emerging (single) Yes Likelihood ratio tests for AE rate (i.e. incorporating exposure time), incorporating recurrent events
Mehrotra & Adewale [41] 2012 HT Emerging (multiple) No P-value adjustment
Mehrotra & Heyse [42] 2004 HT Emerging (multiple) No P-value adjustment
Allignol, Beyersmann & Schmoor [43] 2016 E Emerging (single) No Estimates cumulative incidence function in presence of competing risks
Borkowf [44] 2006 E Emerging (single) No Confidence interval for difference in proportions
Evans & Nitsch [45] 2012 E Emerging (single) No Proportions, incidences, odds ratios etc.
Gong, Tong, Strasak & Fang [46] 2014 E Emerging (single) No Non-parametric estimate for mean cumulative number of recurrent events in presence of competing risks
Hengelbrock, Gillhaus, Kloss & Leverkus [47] 2016 E Emerging (single) No Survival based methods to estimate hazard ratios for recurrent events
Lancar, Kramar & Haie-Meder [48] 1995 E Emerging (single) No Non-parametric estimate for prevalence allowing for recurrent events
Leon-Novelo, Zhou, Nebiyou Bekele & Muller [49] 2010 E Emerging (multiple) No Bayesian approach to estimate the probability of severity grading of events in treatment and control groups separately
Liu, Wang, Liu & Snavely [50] 2006 E Emerging (single) No Confidence interval for difference in exposure adjusted incidence rates
Nishikawa, Tango & Ogawa [51] 2006 E Emerging (single) No Estimates the cumulative incidence function in presence of competing risks and conditional estimate for recurrent events
O’Gorman, Woolson & Jones [52] 1994 E Emerging (single) No Confidence intervals for difference in proportion
Rosenkranz [53] 2006 E Emerging (single) No Survival based method to estimate dependence between AE time and discontinuation time
Siddiqui [15] 2009 E Emerging (single) No Non-parametric estimate for the cumulative mean number of events allowing for recurrent events
Sogliero-Gilbert, Ting, & Zubkoff [54] 1991 E Emerging (multiple) No A score to indicate abnormal laboratory values
Wang & Quartey [55] 2012 E Emerging (single) No Non-parametric estimate for mean cumulative event duration allowing for recurrent events
Wang & Quartey [56] 2013 E Emerging (single) No Semi-parametric estimate for mean cumulative event duration allowing for recurrent events
Berry [57] 1989 DMP Prespecified Yes Bayesian approach to estimate the posterior probability that event rate or incidence rate (incorporating exposure time) is greater in the treatment group compared to control group
French, Thomas & Wang [58] 2012 DMP Prespecified Yes Bayesian logit model and a piecewise exponential models to give posterior probabilities that predefined risk difference threshold is exceeded
Yao, Zhu, Jiang & Xia [59] 2013 DMP Prespecified Yes Bayesian beta-binomial model to give posterior probability that predefined risk difference threshold is exceeded
Zhu, Yao, Xia & Jiang [60] 201638 DMP Prespecified Yes Bayesian gamma-Poisson model to give posterior probability that predefined risk difference (incorporating exposure time) threshold is exceeded
Berry & Berry [61] 2004 DMP Emerging (multiple) No Bayesian hierarchical logit model to give posterior probability that event rate greater in treatment group compared to control group
Chen, Zhao, Qin & Chen [62] 2013 DMP Emerging (multiple) Yes Bayesian hierarchical logit model to give posterior probability that event rate greater in treatment group compared to control group for interim analysis
Gould [63] 2008 DMP Emerging (multiple) No Bayesian approach to estimate the posterior probability that AEs in treatment group produced by a larger process than AEs in control group
Gould [64] 2013 DMP Emerging (multiple) No Bayesian approach to estimate the posterior probability that AEs in treatment group produced by a larger process than AEs in control group accounting for exposure time
McEvoy, Nandy & Tiwari [65] 2013 DMP Emerging (multiple) No Bayesian multivariate approach to give posterior probability of difference in event rates based on indicator functions
Xia, Ma & Carlin [66] 2011 DMP Emerging (multiple) No Bayesian hierarchical logit and log-linear (incorporating exposure time) models to give posterior probability that event rate greater in treatment compared to control group
  1. aV Visual, HT Hypothesis Testing, E Estimation, DMP Decision-Making Probabilities