Concept and Implementation

Data De-Duplication
in DATIM
Concept and Implementation
DATIM
Data for Accountability, Transparency, and Impact
De-Duplication Overview
•
•
•
•
•
Background
De-Duplication in PEPFAR
De-Duplication in DATIM
De-Duplication Logic
Standardizing De-Duplication Efforts
DATIM
Data for Accountability, Transparency, and Impact
2
A Brief History of Duplication
• “The numbers don’t add up”
– PEPFAR totals vs. National results
– Same Things, Same Place, Same Time
• “I did this” vs. “We did this”
– Partners report individual accomplishments
– Programs report aggregate, unique results
• Double-counting vs. Service Overlaps
– Sometimes it’s simple (2 partners, exact same clients)
– Sometimes it’s complex (overlapping community interventions
across populations, geographic areas)
• Program Accountability
DATIM
Data for Accountability, Transparency, and Impact
3
De-Duplication in PEPFAR I & II
• PEPFAR To Date: De-Duplicate Results at
Technical Area Level Only
– Map Partners, Services, Indicators
– Add up results
– Compare to other data sources (MOH national
results, Global Fund reporting, etc.)
– Exclude, extrapolate, estimate
– Report single, OU-level results by indicator with
narrative description of any de-duplication efforts
DATIM
Data for Accountability, Transparency, and Impact
4
De-Duplication in DATIM
• DATIM Going Forward: De-Duplicate Results
at Site Level
– Receive site-level data from IPs
– Review, Approve, Submit data to Interagency
– Interagency (SI, PCO) use Data De-Duplication App to
review duplicates by site/indicator
– Select standard de-duplication logic where applicable
– Address unique/contextual duplication or overlap issues at
site level case by case
– Review and address DSD-TA “crosswalk” overlaps
– Submit to HQ
DATIM
Data for Accountability, Transparency, and Impact
5
Defining De-Duplication Terms
• Pure/Normal/Simple De-Duplication: All DSDDSD or TA-TA duplicate (or differing) values to
be addressed within DATIM
• DSD-TA Crosswalk/Complex De-Duplication:
When multiple partners report DSD and TA
results at the same site, which double-counts
patients/beneficiaries across DSD and TA
results; revise TA value within DATIM
DATIM
Data for Accountability, Transparency, and Impact
6
Standard De-Duplication Logic
• SUM: Add all IM results together at site level
(No Duplication of clients by IMs)
• MAX: Across all IMs, report the “highest reported
value” of each indicator/disaggregation at the site
level
(Full or Broad Duplication of clients by IMs)
• CUSTOM: Across all IMs, report a “custom value”
at the site level based on contextual or known
programmatic factors
(Partial Duplication of clients by IMs)
DATIM
Data for Accountability, Transparency, and Impact
7
Site-Level “Simple”
Duplication Scenarios:
Community Partners
DATIM
Data for Accountability, Transparency, and Impact
8
OVC_SERV Community De-Duplication
Scenario 1: No Duplication (SUM)
DISTRICT X
Total OVC_SERV
Active Beneficiaries
30 + 30 + 30
VILLAGE B
90
(Intervention Point)
PARTNER 2: 30
(Case Management)
VILLAGE A
(Intervention Point)
DATIM
PARTNER 1: 30
(Case Management)
Data for Accountability, Transparency, and Impact
VILLAGE C
(Intervention Point)
PARTNER 3: 30
(Case Management)
OVC_SERV Community De-Duplication
Scenario 2: Broad Duplication (MAX)
DISTRICT X
PARTNER 2: 60
(Psychosocial Support)
Total OVC_SERV
Active Beneficiaries
30 ≤ 60 ≤ 90
VILLAGE B
90
(Intervention Point)
VILLAGE A
(Intervention Point)
DATIM
PARTNER 1: 30
(IGA Partner)
Data for Accountability, Transparency, and Impact
VILLAGE C
(Intervention Point)
PARTNER 3: 90
(Case Management)
OVC_SERV Community De-Duplication
Scenario 3: Full Duplication (MAX)
DISTRICT X
PARTNER 2: 90
(Psychosocial Support)
Total OVC_SERV
Active Beneficiaries
90 ≤ 90 ≤ 90
VILLAGE B
90
(Intervention Point)
VILLAGE A
(Intervention Point)
DATIM
PARTNER 1: 90
(IGA Partner)
Data for Accountability, Transparency, and Impact
VILLAGE C
(Intervention Point)
PARTNER 3: 90
(Case Management)
OVC_SERV Community De-Duplication
Scenario 4: Partial Duplication (CUSTOM)
DISTRICT X
PARTNER 2: 60
(Psychosocial Support)
Total OVC_SERV
Active Beneficiaries
50% Overlap
VILLAGE B
90
(Intervention Point)
VILLAGE A
(Intervention Point)
DATIM
PARTNER 1: 60
(IGA Partner)
Data for Accountability, Transparency, and Impact
VILLAGE C
(Intervention Point)
PARTNER 3: 60
(Education, Child Protection)
Site-Level “Simple”
Duplication Scenarios:
Clinical Facility Partners
DATIM
Data for Accountability, Transparency, and Impact
13
TX_CURR Facility De-Duplication
Scenario 1: Full Duplication (MAX)
SITE X
Total TX_CURR_DSD
Current on Treatment
90 ≤ 90
90
TOTAL NUMBER
OF ART CLIENTS
N = 90
PARTNER 1:
90 Clients
DATIM
Data for Accountability, Transparency, and Impact
PARTNER 2:
90 Clients
TX_CURR Facility De-Duplication
Scenario 2: Broad Duplication (MAX)
SITE X
Total TX_CURR_DSD
Current on Treatment
50 ≤ 90
90
TOTAL NUMBER
OF ART CLIENTS
N = 90
PARTNER 1:
90 Clients
DATIM
Data for Accountability, Transparency, and Impact
PARTNER 2:
50 Clients
TX_CURR Facility De-Duplication
Scenario 3: Partial Duplication (CUSTOM)
SITE X
Total TX_CURR_DSD
Current on Treatment
25% Assumed
Overlap
PARTNER 1:
60 Clients
DATIM
Data for Accountability, Transparency, and Impact
TOTAL NUMBER
OF ART CLIENTS
N = 90
90
PARTNER 2:
60 Clients
TX_CURR Facility De-Duplication
Scenario 4: No Duplication (SUM)
SITE X
Total TX_CURR_DSD
Current on Treatment
45 + 45
90
TOTAL NUMBER
OF ART CLIENTS
N = 90
PARTNER 1:
45 Clients
DATIM
Data for Accountability, Transparency, and Impact
PARTNER 2:
45 Clients
DSD-TA “Crosswalk”
Duplication Scenarios:
Some Illustrations
DATIM
Data for Accountability, Transparency, and Impact
18
General Principles:
1. De-duplicated MER DSD + TA results =
Total unique patients/beneficiaries in country
2. DSD results prioritized over duplicate TA
results for partners reporting the same results
3. In these duplication cases, TA value must be
determined by user (zero? non-zero?)
DATIM
Data for Accountability, Transparency, and Impact
19
TX_CURR_DSD vs. TX_CURR_TA
Scenario 1: Full Duplication (TA=0)
SITE X
TX_CURR_DSD &
TX_CURR_TA
Current on Treatment
90 DSD cancels
out 90 TA
TOTAL NUMBER
OF ART CLIENTS
N = 90
90 DSD
0 TA
(90 total)
PARTNER 1: Service Delivery
90 DSD Clients
Quarterly Monitoring + HCW Salary
DATIM
Data for Accountability, Transparency, and Impact
PARTNER 2: QI for Retention
90 TA Clients
Quarterly Monitoring Only
TX_CURR_DSD vs. TX_CURR_TA
Scenario 2: Duplication, DSD > TA (TA=0)
SITE X
TX_CURR_DSD &
TX_CURR_TA
Current on Treatment
90 DSD cancels
50 TA
TOTAL NUMBER
OF ART CLIENTS
N = 90
90 DSD
0 TA
(90 total)
DATIM
PARTNER 1: Service Delivery
90 DSD Clients
Quarterly Monitoring & ARVs
Data for Accountability, Transparency, and Impact
PARTNER 2: QI for Retention
50 TA Clients
Quarterly Monitoring Only
TX_CURR_DSD vs. TX_CURR_TA
Scenario 3: Duplication, TA > DSD (CUSTOM)
SITE X
TX_CURR_DSD &
TX_CURR_TA
Current on Treatment
TOTAL NUMBER
OF ART CLIENTS
N = 90
PARTNER 1: Service Delivery
50 DSD Clients
Quarterly Monitoring & ARVs
DATIM
Data for Accountability, Transparency, and Impact
50 DSD,
(90 TA – 50 DSD)
= 40 TA
50 DSD
40 TA
(90 total)
PARTNER 2: QI Partner
90 TA Clients
Quarterly Monitoring Only
“Simple” De-Duplication Logic
SUM:
MAX:
CUSTOM:
DATIM
Data for Accountability, Transparency, and Impact
23
DSD-TA Crosswalk De-Duplication
Enter TA value that appropriately
de-duplicates the DSD-TA Crosswalk
Indicator total
DATIM
Data for Accountability, Transparency, and Impact
24
Why Standardize De-Duplication?
• Allows PEPFAR to simplify and normalize
duplication handling globally
• Enforces accountability and auditability for
de-duplication logic/decision-making
• Preserves partner-level data without source
modifications
• DATIM requires standard rules & logic
DATIM
Data for Accountability, Transparency, and Impact
25
What happens inside/outside DATIM?
• Identify “pure” duplicates in DATIM across IMs and agencies
• Review and confer with Activity Managers, TWGs, Partners
• Choose de-duplication approach by policy and/or case by
case
• Remediate “pure” duplicates in DATIM
• Review, confer, & remediate “crosswalk” duplicates in
De-Duplication App
DATIM
Data for Accountability, Transparency, and Impact
26
DATIM
Data De-Duplication App
Live Demonstration
DATIM
Data for Accountability, Transparency, and Impact
27
Questions?
DATIM
Data for Accountability, Transparency, and Impact
28