• The code is the ultimate documentation of how data analysis was done
• Need to be able to regenerate an entire analysis and report with a single command
• Allows others to reproduce your work
• Allows you to easily re-run analyses upon data corrections/updates or changes in statistical analysis
• Team work and personnel changes
• Journals starting to require code

• Requires statistical software to be run, i.e., report not self-contained and useable offline. Think RShiny
• E.g. change the bandwidth and re-run a nonparametric smoother for trend; selection of variables to include in a model

• Zoom, pan
• Rescale axes
• Extra information pop-up (hover text)
• Select which traces to show
• Instead of having legends and explanations (e.g., for box plots) show extra information as hover text

• Implementation of javascript D3 graphics model plotly
• Best developed partially interactive scientific graphics for R
• Has it’s own model, or:
• ggplotly function: pass any ggplot2 graphics object through it to get interactivity

Examples Using Mayo Clinic pbc Dataset

p <- plot(d); p$Categorical

p$Continuous

s <- summaryM(bili + albumin + stage + protime + sex + age + spiders +
              alk.phos + sgot + chol ~ drug, data=pbc,
                            overall=FALSE, test=TRUE)

Plots for Categorical Variables

plot(s, which='categorical')

plot(s, which='continuous', vars=1 : 4)

• Show data in almost full resolution using 100 or 200 bins
• Hovertext to see location and count
• Overall statistical summary by hovering over leftmost part
• Can see bimodality, digit preference, …
• SD and (better) Gini’s mean difference aligned to origin
• Intervals under histogram shows quantiles

getHdata(support2)
with(support2, {
    units(meanbp) <- 'mmHg'
    histboxp(x=meanbp, group=dzgroup, sd=TRUE, bins=200)
} )

require(rms)

f <- npsurv(Surv(fu.yrs, status) ~ spiders, data=pbc)
survplotp(f, time.inc=1, times=c(5, 10), fun=function(y) 1 - y)

html(s, caption='Baseline characteristics by randomized treatment',
     exclude1=TRUE, npct='both', digits=3,
     prmsd=TRUE, brmsd=TRUE, msdsize=mu$smaller2)

• Foster good statistical analysis, graphics, reporting practice
• Minimize programming for individual clinical trials
• Reviewers are tired of tables and reports having 100s of pages
• Clinical trial reports have many standard components

• accrual summary
• patient flow/exclusions
• baseline description
• longitudinal analyses
• adverse events
• lab safety parameters (blood, ekg, etc.)
• event timing/incidence
• sequential monitoring of event probabilities

• Tables do not lead to pattern recognition
• Graphics > tables when >2 numbers in the table
• Graphics should use features humans most accurately perceive
  • position along a common scale
• Need for signposts on graphics
• Tables are secondary; should be in appendix and hyperlinked
  • or as hovertext from a plot

• Emphasize confidence intervals for differences
• Show entire distributions when possible
• Favor quantiles over moments
• Percentages are inherently confusing
  • replace with proportions and ratios

• Utility functions
• High-level report component functions
• Unified handling of figure generation, captions
• New graphical elements

• Extended box plots
• Special dot charts for stratified proportions
• Half-violin plots (vertical density plots)
• Half-confidence intervals
  • centered at midpoint of two estimates
  • length = ½ length of CL
  • → touches the 2 estimates ↔ difference not “significant” at α=0.05
• Spike thermometers as signposts

• accrualReport: subject accrual, accounting for regions, countries, sites
• exReport: exclusion and improper randomization report
• dReport: descriptive stats for baseline and longitudinal data
• eReport: event report
• survReport: time-to-event report

• nriskReport: number–at–risk report (declining denominators for longitudinal data)
• (future): sequential monitoring of event incidence/stopping boundaries
• analysis variables ~ stratification vars + id(subject ID var)

• R + knitr + LaTeX + pdflatex + Acrobat Reader
• Pop-ups provided by javascript show detail
• Report using real RCT data
  Longitudinal data, no geographical regions
• Report using simulated RCT data
  Multinational clinical trial: region/country/site

• Exquisite control of formatting; beautiful printing
• Only Adobe Acrobat Reader supports javascript in pdf files, for pop-ups etc.
• Acrobat Reader is poorly supported and bloated
• Minor update to Acrobat Reader on Macs disabled pop-ups

• Extensive styling/programming in LaTeX is hard
• Copying and pasting advanced tables from pdf into Word doesn’t work well
• Graphics are static, without drill-down
• Code all present/absent

• RMarkdown → html documents
  • html reports allow interactive graphics
• HTML5, self-contained javascript
• Viewable in any browser
• R functions write HTML
  • Regular tabular output, hyperlinks, navigation bars, etc.
  • Advanced tables
    • htmlTable package and Hmisc summaryM

• R programming key: abstract markup, store translations in a central place
  • plain text, HTML, LaTeX
  • Go through the pain of figuring out markup for χ ² ₇ once
  • R Hmisc package markupSpecs list: large number of translations and helper functions
  • Special LaTeX/HTML translation tables for functions
  • Fine tuning: edit one file, markup used by many functions

• Almost no tables
• Hover over a graphical element to see the relevant portion of a table

• HTML file can contain real data, not just relative coordinates of points
• Self-contained HTML files can be large
• No concept of pagination and other special control for pretty printing
• But: Nice format on any device (dynamic resizing)

• Report using real RCT data
  Longitudinal data, no geographical regions
• Report using simulated RCT data
  Multinational clinical trial: region/country/site

• biostat.mc.vanderbilt.edu/RCTGraphics
• github.com/harrelfe

safetyGraphics

• Collaborative open source effort from
  • pharmaceutical industry
  • FDA
  • academia
• Lead developer: Jeremy Wildfire of Rho Inc.
• Extensive clinical input
• First module: hepatotoxicity