MdliModeling Rf lReferral Nt kNetworks to Maternal in Ethiopia...Referral Network Total Population...
Transcript of MdliModeling Rf lReferral Nt kNetworks to Maternal in Ethiopia...Referral Network Total Population...
M d li R f l N t k Modeling Referral Networks to Avert Maternal Death in Ethiopia
Caleb ParkerGIS AnalystBehavioral and Social SciencesFHI
Presentation Overview
• Referral backgroundReferral background• Setting: Ethiopia• Network AnalysisNetwork Analysis
– Objectives, data and methods– Defining referral networksDefining referral networks– Model building and preliminary results
• Next steps in EthiopiaNext steps in Ethiopia
REFERRAL BACKGROUND
What and why referral?
• Defined: upward movement in the seeking of care within p gthe health system.– Emergency Referral
f l i k i h fi ld f l d b• Referral is key in the field of maternal and newborn health– “under‐documented, under‐researched and under‐o , u de e e c ed d u de
theorized.”
• Referral utilizes current resources through linkages in a tsystem
– Cheaper– Short to medium‐term solutions
Time between the beginning of a complication and death
Complication Hours DaysH h Hemorrhage
Postpartum 2
Antepartum 12
Ruptured uterus 1
Eclampsia 2
Obstructed labor 3Infection 6
The 3 Delays Model3 y
Referral has the potential to reduce all 3 delays
DELAY #1 DELAY #2 DELAY #3Recei ing
Referral has the potential to reduce all 3 delays
Deciding to seek care
Reaching a facility
Receiving adequate
care
Onset of Recovery or deathComplication
Pyramidal structure & bypassing
Regional Hospital Receiver
District Hospital
Transport
Health center/post
p
/dispensary
Sender
Adapted from Jahn & De Brouwere, 2001Community
Requisites of a functioning referral systemq gCommunication
F ti iTransport
Functioning referral center
Protocols for senders &receiversreceivers
S M SF P SC M t it f l t i d l i t i C t k l d d f t hSource: Murray SF, Pearson SC. Maternity referral systems in developing countries: Current knowledge and future research needs. Soc Sci & Med 62, 2006.
SETTING: ETHIOPIASETTING: ETHIOPIA
Setting: Ethiopia
• Population: 73 millionp 73
• Area: 1.1 million km2
Twice the size of Texas– Twice the size of Texas
Regions of Ethiopia
hAmhara
Tigray and Amhara – Key Indicators
Tigray Amhara NationalTigray Amhara National
Population (2007 Census) 4,400,000 17,200,000 73,900,000
Number of hospitals 14 18 112Number of hospitals 14 18 112
Percent minimum recommended (CEmOC) 55% 18% 39%
Institutional delivery rate 9% 5% 7%Institutional delivery rate 9% 5% 7%
Cesarean delivery rate 0.7% 0.2% 0.6%
Percent of health centers with desired staffing 45% 2% naPercent of health centers with desired staffing 45% 2% na
OVERVIEW OF DATA AND OVERVIEW OF DATA AND METHODS
Background for this activity
• Ethiopia Needs Assessment for EmONC conductedp
– 797 Healthcare facilities
Geocoded– Geocoded
• Staff saw potential GIS use
• Emergency Referral Network Analysis
– Strengthening referral systems could save lives in short and medium term, even as longer term strategies are implemented
Objectives:
1. Model current referral system y
2. Identify facilities 2+ hours from closest receiving facility
3. Model short and medium term solutions to reduce travel time
Data
1. Ethiopia EmONC Assessmentp2. Population Data – 2007 Census projected to 20083. Digital Elevation Model3 g4. Road Network Data
Ethiopia Elevation
Ethiopia Road Network
Road Network Discrepancies
Road Network Discrepancies
Methods
1 Define Sender and Referral Facilities1. Define Sender and Referral Facilities
2. Model Current Referral Network
3. Model Referral Scenarios
B ild f ilit t h t – Build facility catchment areas
Defining facilities
• Tier A (Receivers)– conducted cesarean delivery in 3 months prior to the assessment
• Tier B (Senders)– all other facilities
• Identified referral networks based on • Identified referral networks based on shortest travel time to a Tier A facility
Model Current Referral Network
Key Variable: Travel Timel l d di l i f h iCalculated direct travel time for each Tier B
facility based on:Di t d t kDistance on road networkRoad surfaceSeason
Calculated adjusted travel time based on:Motorized transportation available on siteCommunication (if no transportation)
Key Variable – Travel Time
Factors Tier B Facility #1 Tier B Facility #2Factors Tier B Facility #1 Tier B Facility #2Direct travel time (min) to closest Tier A
48 118
‐ Road Surface (in GIS) (in GIS)
‐ Season Dry Dry
Transportation on site No (48x2=96) Yes (118x1=118)Communication No (96+30=126) N/AAdj t d t l ti 6 8Adjusted travel time (min)
126 118
Percentage of facilities within 60 and 120 minutes to closest Tier A facilityminutes to closest Tier A facility
C l i i Cl CCumulative Minutes to Closest Tier A Facility
Count Percentage
Under 60 77 35.8%Under 60 77 35.8%61 to 120 64 29.8%More than 120 74 34.4%3
Model building
• Build Catchment AreasM d l 0 St t Q• Model 0: Status Quo
• Model 1: All facilities have motorized transportation and communicationand communication
• Model 2: Upgrading strategic Tier B facilities to Tier A facilitiesfacilities
• Model 3: Improved road surfaces• Optimal model: what is the best outcome with the • Optimal model: what is the best outcome with the
lowest input (additional phones/vehicles)
Creating Catchment Areas
• An area that “should” be served by a facilityAn area that should be served by a facility• Cost Distance Raster in ArcGIS
– SlopeSlope– Rivers/terrain
• Once created, calculated populationOnce created, calculated population
Facility Catchment Areas with Road Network
Referral Networks with Facility Catchment Areas
Referral Networks
Population Density by Catchment Area
Population Density (sq km)
Model building
• Build Catchment Areas• Model 0: Status Quo• Model 1: All facilities have motorized transportation
and communication• Model 2: Upgrading strategic Tier B facilities to Tier A
facilities
M d l 0 STATUS QUO
Facility Catchment Areas Beyond 2-hour Travel Time
Model 0: STATUS QUO Total Population
Population Currentlywith Access
10,019,395 5,626,099 56.2%
Model building
• Build Catchment AreasBuild Catchment Areas• Model 0: Status Quo• Model 1: All facilities have motorized transportation Model 1: All facilities have motorized transportation
/communication• Model 2: Upgrading strategic Tier B facilities to Tier A Model 2: Upgrading strategic Tier B facilities to Tier A
facilities
M d l 0 STATUS QUO
Facility Catchment Areas Beyond 2-hour Travel Time
Model 0: STATUS QUO Total Population
Population Currentlywith Access
10,019,395 5,626,099 56.2%
Facility Catchment Areas Beyond 2-hour Travel Time
Model 1: Increase in population served by ensuring that every Tier B facility has a motorized vehicle
Total Population Currently Population with Overall Increase in Referral Network Population with Access* Access in Model 1 Population Access
1412 – Pawe 1,338,664 646,721 48.3% 646,721 48.3% ‐ 0.0%
1505 – Debre Tabor 2 723 313 2 182 465 80 1% 2 447 996 89 9% 265 531 9 8%1505 Debre Tabor 2,723,313 2,182,465 80.1% 2,447,996 89.9% 265,531 9.8%
1543 – Felege Hiwot 2,156,109 1,151,324 53.4% 1,994,981 92.5% 843,657 39.1%
1623 – Debre Markos 3,801,309 1,645,589 43.3% 2,660,041 70.0% 1,014,452 26.7%
TOTAL 10,019,395 5,626,099 56.2% 7,749,739 77.3% +2,123,640 21.2%
*Only for persons at the Tier B facility
Facility institutional delivery rates
Institutional delivery rate
M d l 0 STATUS QUO
Facility Catchment Areas Beyond 2-hour Travel Time
Model 0: STATUS QUO
Facility Catchment Areas Beyond 2-hour Travel Time
Model 2: Increase in population served by upgrading Fenote Selam District Hospital to a Tier A
Total Tier B Facilities in New Tier A
Referral Network
Total Population in New Referral Network
Population with Access to their Current Tier A
Population with Access to New Tier A
Overall Increase in Population with
Access
16 1 945 023 293 895 15 1% 781 640 40 2% 487 745 25 1%16 1,945,023 293,895 15.1% 781,640 40.2% +487,745 25.1%
Next Steps in Ethiopia
Further modelingg• Combinations of models (ex. Motorized vehicles plus upgrading)• Improved road surfaces• Optimal model: what is the best outcome with the lowest input • Optimal model: what is the best outcome with the lowest input
(additional phones/vehicles)• Costing
Federal MoHWorking Group on Referral• Site selection for new ambulancesSite selection for new ambulances• Testing new management models• Transfer of methods and validation of assumptions