How to make HTML reports acceptable to a medical audience

In February I had posted on the same subject here, but there were no replies and the post is closed now:

After seeing a video by Frank Harrell at RStudio 2020, I started a similar thread here.

In short: I am looking for a tool that enables reviewing similar to what you can do with Word or PDF. Easy with servers (e.g. using Hypothesis or Zoho), but servers are not acceptable in a clinical environment for security and maintenance reasons.

My two cents,

For R Markdown based documents that need reviews, I often use github.
For example, with a new blog post, that blog post will appear as a pull request on a blogdown site repo (in your case, your bookdown repo), reviewers are added, and are able to add comments. These reviews can be line item or comments on the overall PR. Here's some documentation on that workflow, Features · Code review · GitHub.
This strikes me as the best solution for a bookdown repo.

I do work with people who are not comfortable with git or github, and cannot be convinced to become familiar. In those situation I tend to turn to google-docs, similar to your word doc workflow.

It's not clear to me that emailing around attached annotated word documents is any more secure than another web-based option with appropriate authentication. But emailing word docs doesn't strike me as that bad of an option given your constraints and the preferences of your audience.

I'd be curious if there's a good option? Seems like there should be a chrome extension that let's you add comments, and then same the page as a single file.

It's not clear to me that emailing around attached annotated word documents is any more secure than another web-based option with appropriate authentication.

That's correct, but reality strikes back: Emailing is allowed in a clinical context, installing a server or using github is not possible because then the responsibility is on the side of the IT department, and they will block any changes. Surprisingly, Email is never encrypted, and I have to fight that at least data are zip-encrypted ("Why? We never did this"). Don't ask me about the logic behind this.

But emailing word docs doesn't strike me as that bad of an option...

Frank Harrell in his RStudio talk advocated HTML documents because they support interactivity (DT, plotly), and that's lost after conversion.

The ideal would be a JS package that can be added (nicknamed AnnotDown to honor the other downs) into a report. Installing browser extensions is already difficult, because browser installations are locked in clinical context.

The bottleneck is saving the changes - creating a compound document after the edits with a desktop program could be considered too complicated - "it works with Word and PDF, why should we..."

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.