Table with characteristics of trials included in dashboard

[delwen]

Table with overview of characteristics of trials included in the dashboard. We'll probably include it in a separate "Trials" page. @maia-sh: assigning you since you already made a first version, but let us know if we can support with this!

[maia-sh]

To confirm: the dataset to use for the characteristics table is ct-dashboard-intovalue-all.csv?

[maia-sh]

I've created a trial characteristics table with a broad set of characteristics: https://github.com/quest-bih/clinical-dashboard/blob/characteristics-table/create-characteristics-table.R

Some open questions:

• is this the set of characteristics we're interested in? i didn't include outcome measures since those are in the plots, just "table 1"-esque elements
• what dates do we want? registration, start, completion? As years or range? Right now, there are both for show
• as usual, not sure how to address the UMC-led vs. industry sponsored so playing it safe with the wording for now, and we can adjust as needed.
• center size is pretty hackey from intovalue so just used as is and included the explanation in the caption. I don't think it's worth extra work to refine, so either we use as is or drop
• i included euctr but not as "cross-reg" since not manually checked and just as "EUCTR Trial ID in Registration" (we will check charite though for track value)
• @bgcarlisle does this format work for you in terms of display? anything else i should tweak format-wise?

I won't be working in this branch, so alternatively feel free to check out and work on it there.

[delwen]

Thanks @maia-sh!

Some first thoughts:

• I think it makes sense to build the table on the ct-dashboard-intovalue-all.csv dataset in a first pass. Showing UMC-specific numbers (ct-dashboard-intovalue-umc.csv) would probably be more informative for individual UMCs, but it would quickly get overwhelming/non-readable and we'd probably have to lose the registry stratification. TBD
• Agree to not include outcome measures as these are already in the plots
• Dates: I find registration and completion most informative, as years. Registration could still a bit confusing though, with some trials registered >2017 even though we filter for trials completed until 2017. So maybe completion only?
• UMC-led vs sponsor: I definitely think we should include the number of trials with an industry sponsor. Suggest to include a sentence in the same page as the table (or in caption) that again clarifies how trials were associated to UMCs. We could either list the fields used in CT.gov/DRKS, or explain it in a more narrative format. In short, explain why industry-sponsored trials show up in this UMC dataset.
• Center size: no strong opinion but I'm tempted to leave this out unless we can find a more commonly used definition
• EUCTR: looks good, and agree to describe it as "EUCTR Trial ID in Registration"
• We can remove "A trial was considered prospectively registered if registered in the same or previous months to start date." from the caption

[delwen]

We could also think of including Table 1b with characteristics of trials included for prospective reg in CT.gov. Same info but just for this updated dataset. Would make it clearer that this metric comes from a different dataset, and I'm guessing there would be interest re what types of trials are included in this metric

[maia-sh]

• UMC-specific tables: don't think that makes sense on this first pass; if we see interest I guess we could try some sort of toggled table
• Dates: trials completed by 2018 NOT 2017. Fine with any of the 3 dates. Not making changes for now.
• Center size: leaving for now until we discuss
• UMC-led vs sponsor: I will add to caption for now. https://github.com/quest-bih/clinical-dashboard/commit/1be4711d9d1cfc759153fc6f630193664e6000ce
• Removed the prospective reg line from caption
• Table 1b for prospective reg: i don't think we NEED to but sure can add (except center size which I take from intovalue and not from registry). Will hold off until we discuss

[maia-sh]

• Center size: remove
• Date: start and completion as categorical (remove registration)
• UMC-specific table: not for now; @bgcarlisle can make a toggle later if wanted
• Table 1b for prospective reg: yes add @bgcarlisle

@bgcarlisle will make a new page for the 2 tables

[delwen]

Just a thought for the manuscript - we could also add to the characteristics table (or create a new table) with the numbers for our different denominators. So we could start with the overall number of trials, then the number of trials with a publication and PMID (TRN abs/linkage in registry), the number of trials with a publication and full text (TRN ft), etc. Again, stratified by registry. Would drive home the point that we have different denoms for most metrics and how the numbers change.

In the paper I already made a table that gives an overview of the practices, the rationale for inclusion, and the sample of trials in each case. Showing the denoms could complement that quite well.

@maia-sh @bgcarlisle thoughts?

[delwen]

@bgcarlisle anything we still need to do/discuss for the integration of the trial characteristics table (for IV sample and the CT.gov prospective reg sample) in the dashboard?

[bgcarlisle]

Okay, I've integrated the table into the dashboard!

[delwen]

@bgcarlisle: thanks! We also need to add the same table for the CT.gov prospective reg sample.

[bgcarlisle]

We don't have the same columns in the prospective registration data, so we can't generate the same table, unless we get those data from ct dot gov

@maia-sh We need a bunch of other columns like is_randomized, is_multicentric, etc that aren't currently in the prospective registration CSV; is this much trouble to grab?

[maia-sh]

i can do that

[maia-sh]

I ended up functionalizing some AACT things, which took a bit longer, but good to have for future trial id sets. Please make sure to pull from main before any work with intovalue-data repo! The updated dataset should have the variables you @bgcarlisle need to make the characteristics table: https://github.com/maia-sh/intovalue-data/blob/ctgov-2018/data/ctgov-2018/prospective-reg-ctgov-2018-trials.csv I left all the variables in this time for completeness. phase_unified, start_year, and industry_sponsor you should create through the same pre-processing you do on iv_all in https://github.com/quest-bih/clinical-dashboard/blob/main/trial_characteristics_page.R Let me know if you get any snags!

[bgcarlisle]

This is almost perfect! This just appears to be missing the cities column

[maia-sh]

try now!

[bgcarlisle]

Amazing!!

The new table is up now!

https://github.com/quest-bih/clinical-dashboard/commit/f4d4341f5ea672bbf4323a98699b4d75715b5bfc

[maia-sh]

Great! Closes #26