Designing Historical Analysis

[tbhallett]

This PR does the analysis to estimate the impact of health of the increases in Human Resources for Health (healthcare workers) have occurred already in Malawi.

Dominic Nkhoma has provided data on the number of staff (4th July 2024), saved in the team Dropbox.

Basic descriptions of these data are shown below (done in the script src/scripts/data_file_processing/healthsystem/human_resources/analysis_historic_changes_in_hr.py)

[BinglingICL]

Nice trend before 2021. But big drops from 2021 to 2023, overall and consistently across all districts (except Likoma seemly). Was it because Covid-19 relevant or funds declined?

[tbhallett]

Nice trend before 2021. But big drops from 2021 to 2023, overall and consistently across all districts (except Likoma seemly). Was it because Covid-19 relevant or funds declined?

I asked Dominic and he replies,
"""Dear Tim
The 2023 figures are wrong. I will investigate, for now please exclude 2023. Otherwise the rest looks good.

regards

Dominic
"""

So I'll do that for now.

[tbhallett]

Construction of scenarios

[tbhallett]

First run of src/scripts/impact_of_historical_changes_in_hr/scenario_historical_changes_in_hr.py

job_id = historical_changes_in_hr-2024-07-08T145329Z

Figures are

[tbhallett]

So, the analysis and plotting script basically work and give some sensible-seeming results.

Things to do following discussion with team:

• Improve plotting: Show breakdown by cause of deaths/dalys;
• Assemble summary slide deck
• Look at squeeze-factor over time in calibration plots (from Mode 1): Does it show a trend over time consistent with there being more HCW available (i.e., does the squeeze reduce over time)
• Test whether adjustment to the scaling calculation [to take account of the effect that we might be over-correcting?] make a difference to the 'delta' between the scenarios.
• Query numbers with Dominic
• Set priority policy to be the HBP (?)
• Re-run when confirmed analysis design per the abvove