UNIVERSITY OF WASHINGTON Maternal Mortality Estimates: Global Progress on Levels and Trends August 30, 2010 Rafael Lozano, MD MSc Professor of Global Health
UNIVERSITY OF WASHINGTON
Maternal Mortality Estimates: Global Progress on Levels and Trends
August 30, 2010
Rafael Lozano, MD MSc
Professor of Global Health
Outline• Background
• Data sources and methods
• Results
• Lessons learned and next steps
2
Broader context• Important momentum for new estimates
o Need to assess progress given international commitment to Millennium Development Goal 5.
• Opportunities for robust reassessmento WHO sponsored development of country estimates for 1990, 1995, 2000,
2005.
o Most recent assessment (Hill et al, 2007) reported 576,300 maternal deaths in 1990 and 535,900 in 2007; 0.48% annual rate of decline in the global maternal mortality ratio.
• Methodological advances
• More data than ever before
3
Outline• Background
• Data sources and methods
• Results
• Lessons learned and next steps
4
Definitions• Maternal death: “death of a woman while pregnant or
within 42 days of termination of pregnancy…from any cause related to the pregnancy or its management, but not from accidental or incidental causes” (ICD-10)
• For inter-temporal and international comparisons, ICD and MDG manual recommend A+B+C excluding D+E+F
• Maternal mortality ratio: number of maternal deaths per 100,000 live births
Direct Indirect HIV Incidental
Early maternal (<42 days) A B C D
Late maternal (> 42 days & < 1 year) E F
5
Density of site-years of observation, 1980-2008
6
Systematic identification of data• Vital registration data
o Naghavi et al.* have undertaken a correction of the registration data, increasing maternal deaths in vital registration countries by 40% on average.
• Sibling history datao Survey respondents report on survival of siblings
and whether deaths of sisters 15-49 occurred during pregnancy, childbirth, or period after birth.
o Apply the Gakidou-King correction for survival bias.
• Deaths in the household
• National and sub-national verbal autopsyo Literature review to identify published estimates
from national and sub-national population-based studies of maternal mortality.
7
* Naghavi M, et al. Algorithms for enhancing public health utility of cause s of death data, Population Health Metrics, 2010.
Source of DataSite-Years of Observation
Vital registration 2,186Sibling Histories 204
Surveillance Systems 20Census/Survey Deaths in
Household 26National VA 35
Subnational VA 180Total 2,651
No data for 21 countries, representing 2.2% of births
Analytical approach
• Extract the proportion of all female deaths due to maternal causes for five-year age groups (15-49).
• Apply proportion to new adult female mortality time series (Rajaratnam et al, 2010) to arrive at maternal mortality rates by country, year, and age groups.
• Two-stage analysis, modeling the age-specific maternal mortality rate:o Stage 1: Linear model
o Stage 2: Spatial-temporal model
• HIV counterfactual analysis
• Model validation
8
Criteria for better and consistent estimates
• Continued increase in quantity and quality of data
• Use of models with improved predictive validity
• Quantify uncertainty using statistical methods
• Transparency of data sources and methods
9
Outline• Background
• Data sources and methods
• Results
• Lessons learned and next steps
10
Maternal deaths by region, 1980-2008
11
Regional trends in the MMR
12
13
MMR with and without HIV
India: Predicted MMR per 100,000 live births with uncertainty
14
MMR per 100,000 live births, 2008
15
Annualized rate of decline in MMR, 1990 to 2008
16
Outline• Background
• Data sources and methods
• Results
• Lessons learned and next steps
17
Lessons learned• Gaining ground
o Global maternal deaths down to 342,900 in 2008
o Global trend is a 1.4% decline per year since 1990
o 23 countries are on track to meet MDG 5, achieving an annual rate of decline of 5.5%
• Progress undocumented but not unexpectedo Global total fertility rate has dropped from 3.70 in 1980 to 2.56 in 2008
o Income per capita has been rising over the period, particularly in Asia and Latin America
o Maternal education has been increasing as well; in Sub-Saharan Africa, the average years of schooling for women aged 25-44 rose from 1.5 years in 1980 to 4.4 years in 2008
18
Lessons learned• Adverse Impact of HIV
o Progress on reducing maternal mortality would have been much greater in the absence of HIV, especially in sub-Saharan Africa
o Important implications for intervention policy
• Communication with countries and local researcherso Letters to the editor: Argentina. Jamaica, Iran, Burkina Faso, Peru,
Philippines, Indonesia
o Letters to correspondence author: Turkey, Jamaica, Centro America, Cambodia
19
Next steps
• Research o Continue updates
o Evaluation
• Data disseminationo Reports
o Web page http://www.healthmetricsandevaluation.org
• Organization of regional workshops
20
KNOWING IS NOT ENOUGH; WE MUST APPLY.
WILLING IS NOT ENOUGH; WE MUST DO.
JOHANN W. VON GOETHE (1749-1832)
21