right sizing net new

PEPFAR systems do not account for site shifts or close out when calculating TX_NET_NEW, which can give grossly over/under-inflated values that are not representative of what is actually occurring. This presentation provide an alternative approach to calculating TX_NET_NEW

Transcript

1. right sizing net new cop 20 | tanzania
2. the problem
3. pepfar systems do not account for site shifts or close out when calculating net new, which can give grossly over/under-inflated values that are not representative of what is actually occuring tx_net_new = tx_curr – tx_currprior pd
4. a thought exercise
5. the leo clinic in callisto is one of pepfar/jupiter’s sites reporting on current on treatment (tx_curr) leo clinic
6. orion, an usaid partner, was reporting treatment patients in at the leo clinic the last two quarters leo clinic tx_curr fy2050q3 fy2050q4 fy2051q1 orion 100 120
7. orion’s net new between q3 and q4 was 20 leo clinic tx_curr fy2050q3 fy2050q4 orion 100 120 tx_net_new orion 20
8. this year, the site transition to a cdc parnter, virgo, who started reporting treatment at the clinic leo clinic tx_curr fy2050q3 fy2050q4 fy2051q1 orion 100 120 virgo 130
9. in the msd and panorama, it would like orion lost 120 people on treatment and virgo found 130 leo clinic tx_curr fy2050q3 fy2050q4 fy2051q1 orion 100 120 virgo 130 tx_net_new orion 20 -120 virgo 130
10. in reality, orion left the site and virgo added only 10 new patients leo clinic tx_curr fy2050q3 fy2050q4 fy2051q1 orion 100 120 virgo 130 tx_net_new orion 20 -120 virgo 130 tx_net_new orion 20 - virgo 10 unadjusted reassigned
11. the method: to calculate net new, the agency and partner meta data are stripped from the site. net new is calculated across time regardless of who works in the site. only at the end does the meta data on partners get added back to each site and attribution take place
12. by leaving the new new unadjusted, this not only has implications not just at the site level, but rolling up to the agency level usaid cdc fy2050 fy2051 leo leo
13. the unadjusted level over inflates one agency’s net new gain and the other’s loss usaid cdc fy2050 fy2051 leo leo fy2051 unadjusted net new jupiter usaid cdc
14. the real situation on the ground is better reflected by looking at what happened in the site during the prior quarter, agnostic to what partner/agency was there and reassigning it to the current partner usaid cdc fy2050 fy2051 leo leo fy2051 unadjusted net new usaid cdc fy2051 reassigned net new jupiter usaid cdc
15. in practice
16. actual pepar data/visuals redacted
17. analysis and visualization code can be found on github github.com/achafetz/COP20/blob/master /R/TZA_NN_Adjustment.R
18. right sizing net new cop 20 | tanzania