Refactor pf desc stats measure

[milanwiedemann]

While reviewing the PR I noticed that there were still a couple things that we should change, here are my suggestions. The main issues and changes I tried to address are:

Main change

• Explicit separation of events frames (clinical events and medications) linked by consultation ID or date, but not both
  • Events linked by date now have an explicit where condition removing events that are matched by consultation id: where(selected_events.consultation_id.is_not_in(pf_ids))

Smaller changes include:

• Reduce repetition of code:
  • Define measures interval in measures.define_defaults()
  • Created two filtered event frames (selected_events, selected_medications) for the measures interval at the top of the script
• Improve consistency of variable names
  • Replaced all occurances of the word pathways with conditions)
  • Consistent naming of filtered event frames by Pharmacy First consultation ID (_pfid_) or Pharmacy First consultation date (_pfdate_)
• Clear structure throughout the script now follows (1) PF consultations, (2) PF conditions, (3) PF medications, and (4) Measures

[viv3ckj]

This does read much clearer. Thanks!