Pharma table types/terminology

[NNaikp]

While working on our internal TFL functionalities I found it difficult to actually name the different types of tables we often do in relation with a CSR. Everybody seems, to know exactly what mean-, box- or violin plots are, but the same does not yield exactly for the tables we do in pharma.

A lot of the table packages out there provides a great language for talking about the different parts of a table, but there are multiple kinds of summary tables and descriptive tables.

Is this something others have faced as well?

For now we have invented our own definitions of a few tables in order to make functions that successfully informs the user of which kind of table they are making simply from the name.

[elimillera]

Hey @NNaikp
We have found a similar distinction. I haven't seen the Within-column summary before however Tplyr has control over how cells are displayed with respect to different summaries. Could you describe the differences between the two and when you would use a within column summary?

[NNaikp]

Hi @elimillera
We use the within-column summary tables for all overview tables such as demographic summaries or summaries over AEs, like the below:

Basically everywhere, where you count by values within a column and want to represent multiple such counts in the output.
Which corresponds to adding multiple layers in Tplyr if I am not mistaken.

---

The cross-column summary is used for something like AE outputs by SOC and PT:

---

It looks like it depends on which functionality is built into your table package. Our internal table package is for styling only. We have a data retrieval function which looks up all of the possible outputs you can create using that dataset based on the data model i.e. CDISC + internal stuff. Hence our approach is more on a frequent output-type level and we create types of outputs that you can alter. I guess other companies do that as well on top of rtables or Tplyr to create a new functions that you call directly in the output generation?

Our workflow is at least like this where the same object can render multiple types of output (even plots when needed):
``

Creating something with ADAE

adae <- NNData(db$adam("adae"),
popfilter = SAFFL == "Y",
infilter = AESER == "Y")

swiftTable(adae, type = "wc-summary")
swiftTable(adae, type = "cc-summary")

Creating something with ADLB

adlb <- NNData(db$adam("adlb"),
popfilter = FASFL == "Y",
infilter = ANELFL == "Y" & ANL01FL == "Y" & TOPICCD == "HBA1C_BLOOD")

meanPlot(adlb)
boxPlot(adlb)
swiftTable(adlb, type = "descriptive")

``

The reason for this post was also to hear what terminology other's were using about these kind of tables if we ever were to have something common :)

Would it be correct to say that in Tplyr terms within-column summaries roughly correspond to multi-layer counts with single by columns? And cross-column summaries correspond to single-layer counts with multiple by columns?