lives saved tool (list): how to create a sub-national projection ingrid friberg, phd
TRANSCRIPT
Lives Saved Tool (LiST):How to create a sub-national
projection
Ingrid Friberg, PhD
Why create a sub-national projection?
• Demography is the basis of all child survival data– Birth counts for baseline
• Noticeably different than national information– i.e India vs. Kerala
• Separate urban and rural
• To use over and over again
SADHS, 1998
What demographic data is needed?
Population by age group and gender Total fertility rate
Life expectancy by gender Age specific fertility rate
Life table Sex ratio at birth
International (sub-national) migration
Edit – Demography (DemProj)
How to create a sub-national projection?
• Easy– Assume national trends are essentially correct– Multiply to get the proportion of the country
• Hard– Assume nothing– Identify regional data and enter it
Create a sub-national projection: the easy way
• Create a national projection – Use EasyProj (population, fertility, mortality)
• Generate the sub-national population– Multiply the first year population by the
percent of the total in the sub-national area
• Adjust the migration values
• Save for reuse
1) Select entire population2) Click multiply3) Enter the percent of the total in the sub-population
National migrant numbers need to be adjusted to reflect sub-population changes.
Limitations and Strengths
Limitations
• Can’t change the time frame (yet)
• May deliver inaccurate births counts
• If using AIM, need to adjust this as well
Strengths
• Easy• Can reuse the same time
frame over and over• Reasonably accurate
Creation of sub-national projection: the hard way
• Collect data for a specific baseline year + trend
• Create a national projection – Do NOT use EasyProj
• Enter the sub-national data• Save this projection
– Can modify underlying tables to reflect the new data (optional)
Population by age group and gender Total fertility rate
Life expectancy by gender Age specific fertility rate
Life table for mortality Sex ratio at birth
International (sub-national) migration
Where is the data from?
• Census data
• National surveys
• Sub-national surveys
• DHS
Limitations and Strengths
Limitations
• Time consuming• May need to estimate
some parameters• If the data is wrong, the
projection is wrong
Strengths
• Very accurate• Can reuse with all time
periods of interest (if you change underlying tables)
Other options
• Combining the easy and the hard way– Run EasyProj– Use multiply to get population information
(if you don’t have it)
– Enter all known available data
Example:South Africa – Eastern Cape
• Start with AIM– Runs 1970-2015– DemProj already run– Altered all numbers to percents
• Adult ART, child ART, cotrimoxazole
• Edit DemProj
SADHS, 1998
Urban Non-urban
12.4% 12.6%
25.0 22.7
27.2 22.2
19.5 19.0
11.7 14.1
4.0 6.4
0.2 3.1
Data Collected• 1st year pop - 2001 Census
• TFR - SADHS 1998
• ASFR - Default
• SexRatio - Default
• Life Expectancy - Health Systems Trust
• Life Table - Default
• Migration - State of SA population report
Example:South Africa – Eastern Cape
• Start with AIM– Runs 1970-2015– Altered all numbers to percents
• Adult ART, child ART, cotrimoxazole
• Edit DemProj– Population, Migration, TFR, Life Expectancy
• Add LiST– Adjust mortality rates to reflect HIV– Edit child survival data and coverage data– Increase facility delivery from 74-96%
Scaling up facility based births from 74% to 95%prevents
~6% of all deaths and ~18% of neonatal deaths
Example – Brazil Rio Grande do Sul
• New Projection– Run EasyProj (1996-2015)
• Edit Projection– Data from 1996 Census, 1996 DHS survey
• Population 1996, TFR, Life expectancy…
• Add LiST– Adjust NMR (11), IMR (19), U5MR (22)– Adjust other data…
Results
• Kangaroo Mother Care (0-70%)– Prevents 95 deaths
• Preventive Postnatal Care (0-70%)– Prevents 290 deaths
• Pneumococcal vaccine (0-99%)– Prevents 128 deaths
• Conclusion: preventive postnatal care might be a good area to invest in here
More information?
• DemProj manual– English, French, Russian, Spanish– www.healthpolicyinitiative.com/index.cfm?id=
software&get=Spectrum– www.healthpolicyinitiative.com– www.futuresinstitute.org/pages/spectrum.aspx
• Local demographer