What test data updates will we need in {pharmaversesdtm} for {admiralpeds}? See https://github.com/pharmaverse/pharmaversesdtm/issues/88

[rossfarrugia]

Example: ensuring subjects in DM exist with different pediatric ages.

[bms63]

maybe a separate data package should be created?

Can also just make data in the package and move out once things are more stable.

[ngiangre]

Whichever solution allows for developing functions and tests faster, let's try that. If the data can be made available fast in {pharmaversesdtm}, then I'd go with putting data there. If not, add data to {admiralpeds} then move out. Probably the latter will be faster, is that a correct assumption?

[manciniedoardo]

@ngiangre you are correct, adding in {admiralpeds} is faster in the short term as you keep all development within one repo. However it creates slightly more work in the medium/long-term as you will have to move the data to {pharmaversesdtm} anyway prior to your first release.

@zdz2101 feel free to make a call here in one direction or the other. At the end of the day both approaches will work, and the {admiral} team itself is split as to which path to pick, so we can leave to your discretion so that you select what works best for your pediatrics team. @kaz462 and I can help support from the {pharmaversesdtm} side when the time comes.

[zdz2101]

I think we can go ahead and proceed with an internal solution first then before release do the proper migration, but if anything, we can also look into Non-vignette articles where one of the use cases is: "Use of a package you don’t want to formally depend on. In vignettes and examples, it’s forbidden to show your package working with a package that you don’t list in DESCRIPTION, e.g. in Imports or Suggests."

[CDC-DNPAO]

I'd suggest using some of the data already on the CDC or WHO sites for test data. For example, #8 at https://www.cdc.gov/growthcharts/percentile_data_files.htm
gives various percentiles (starting with 'P') for BMI-for-age. Similar percentiles are given for weight- and height-for-age. I'd suggest using the 5th and 95th percentiles to test the CDC function. For example, if you give the function a BMI of 19.2789 for a boy (sex=1) 24.5 months of age, the output should say that this child's BMI is at the 95th percentile. I'd suggest using the 5th and 95th percentiles for test data.

BMI values >95th percentile in the CDC growth charts are calculated differently from those below the 95th percentile. For test data for these high BMI values, the BMII percentiles are at
https://www.cdc.gov/growthcharts/extended-bmi-data-files.htm
If you click on the 'selected percentiles....', you'll get another excel file. I suggest you focus on P99_99 (the 99.99th percentile) for testing. One thing to note is that IMO, the guy who created this excel file didn't save enough digits after the decimal point, so the calculated percentiles and z-scores will be slightly different (~4th or 5th decimal place) that the output from cdcanthro. You might also consider downloading the 'sigma' column (column F) so that you can reproduce the values on this web page exactly.

The WHO data are arranged somewhat differently, but if you go to
https://www.who.int/tools/child-growth-standards/standards/weight-for-age
and click on the expanded tables for boys' percentiles at the bottom, you'll see an excel file with various percentiles (P). For WHO, I'd suggest using the P01 (the 0.1st percentile) 5th, 95, and P999 (the 99.9th percentile as test data). WHO calculates z-scores outside -3 and +3 (the 0.14 and 99.86 percentiles) differently than those inside this range.

[CDC-DNPAO]

Oops - you might also consider downloading the 'z-score' expanded tables at the WHO website under the 'Expanded Tables'. The give the weight, height, or BMI values at -4 or +4 z-scores (labelled SD4neg and SD4). This would be a better test of extreme values in the whoanthro than the P01 and P999 percentiles I mentioned above.

[rossfarrugia]

Closing out this discussion now as we have actioned creating our initial test data, and we can incrementally improve these later as we see needed.