Statistical Analysis Plan

This is a statistical analysis plan for our approved trial about gastric emptying in elders and adults. We cannot be 100% precise because some transformation may be needed after data collection, but we will reuse some code from our previous similar work published about gastric emptying during labor.

We do not intend to perform interim analysis.

Needed library

We will need openxlsx (and an extra read.xlsx2 function written by me) to load study data. We will need ez for ezANOVA (repeated measures ANOVA). We will use plyr to help making a summary vector to build graphics. We will use ggplot2 to make graphics (because they are one of the best nowdays). It’s possible that we use additional packages.

library(openxlsx)
library(plyr)
library(ez)
## Registered S3 methods overwritten by 'lme4':
##   method                          from
##   cooks.distance.influence.merMod car 
##   influence.merMod                car 
##   dfbeta.influence.merMod         car 
##   dfbetas.influence.merMod        car
library(ggplot2)
read.xlsx2 = function(bdata){
    bdata=read.xlsx(bdata)
    nv=length(bdata)
    ischar=rep(FALSE,nv)
    for(i in c(1:nv)){
        if(is.character(bdata[,i][0])){
            bdata[,i]=factor(bdata[,i])
        }
    }
    bdata
}

Data structure

We will have two file datasets. One for repeated measurements an anoter for non-repeated variables.

Repeated measurements data structure:

  • id - unique patient identifier
  • t - time label (before, 5, 30, 60, 90 and 120 minutes after study drink)
  • d1 and d2 - antral diameters measured using ultrasound
  • ASTAG - antral area (pi times d1 times d2 / 4)
  • volume - 27 + 14.6 x ASTAG -1.28 x age
  • Each line will be used for a unique mesurement (unique combination of id and t)

Non-repeated measures data structure:

  • id - unique patient identifier
  • age in years
  • group - identifier of elder or young group
  • sex
  • weight in kg
  • height in cm
  • ASA physical status

Data merge

We will use merge(data1,data2,by=“id”) or equivalent to merge data.

After merging, we will create a dummy variable called volume.ml.kg calculated as: volume/weight.

Tables

One table will describe study groups mean(sd) for age, weight and height, and count (proportion) for sex and ASA physical status.

Another table will describe mean(sd) for ASTAG, estimated gastric residual volume in ml and in ml/kg by time;(maybe we will split in more than one table).

Another table will describe RM-ANOVA’s coefficients and p-values.

Hypothesis testing

Wilcoxon-Mann-Whitney tests will be used to compare weight, height and age between groups.

Fisher’s exact test will be used to compare sex and ASA physical status between groups.

We will compute Bayes Factors for all comparisons.

Repeated measures ANOVA and Sphericity assessment / corrections will be performed using ezANOVA from ez package.

Parameters: dv=volume, wid=id, between=group, within=t

Graphics

We will provide a study flow chart

We will plot errorbars of ASTAG x time by study group using ggplot.

Additional graphics may be plotted