THE WORLD BANK IMPROVING NATIONAL CAPACITY TO TRACK MATERNAL MORTALITY TOWARDS THE ATTAINMENT OF THE MDG5 Report on a World Bank/UNICEF/WHO/UNFPA Workshop December 7-8, 2010 (English session) December 9-10, 2010 (French session) Nairobi, Kenya
THE WORLD BANK
IMPROVING NATIONAL CAPACITY TO TRACK MATERNAL MORTALITY TOWARDS
THE ATTAINMENT OF THE MDG5
Report on a World Bank/UNICEF/WHO/UNFPA Workshop December 7-8, 2010 (English session) December 9-10, 2010 (French session)
Nairobi, Kenya
ii
CONTENTS
ACKNOWLEDGEMENTS ...................................................................................................................................... III
ACRONYMS AND ABBREVIATIONS .....................................................................................................................IV
INTRODUCTION................................................................................................................................................... 1
1.1 OBJECTIVES AND EXPECTED OUTCOMES........................................................................................................2
1.1.1 Specific objectives ...............................................................................................................................2
1.1.2 Expected outcomes..............................................................................................................................2 1.2 OPENING SESSION .............................................................................................................................................2
2 PRESENTATIONS BY TECHNICAL EXPERTS................................................................................................... 3
2.1 SEXUAL AND REPRODUCTIVE HEALTH AND THE MDGS: WHO/UNICEF/UNFPA/WORLD BANK (H4) MATERNAL MORTALITY
ESTIMATES – OVERVIEW, HISTORY ..................................................................................................................................3 2.1.1 Discussion and observations ...............................................................................................................4
2.2 METHODS FOR MEASURING MATERNAL MORTALITY................................................................................................4
2.3 GLOBAL MATERNAL MORTALITY ESTIMATES AND TRENDS, 1990-2008 ........................................................................5
2.3.1 Discussion and observations ...............................................................................................................5 2.4 DEMONSTRATION OF PROPOSED MATERNAL MORTALITY ESTIMATION DASHBOARD ......................................................6
2.4.1 Discussions and observations..............................................................................................................6
3 GROUP SESSIONS AND THE WAY FORWARD.............................................................................................. 6
3.1 HOUSEHOLD SURVEY GROUP ...............................................................................................................................6
3.1.1 Presentation: Estimating Maternal or Pregnancy-Related Mortality Using Household Surveys ........6
3.1.2 Discussions and practice .....................................................................................................................7
3.2 CENSUS GROUP ................................................................................................................................................7 3.2.1 Presentation: Estimating Pregnancy-Related Mortality from the Census...........................................7
3.2.2 Discussions and Practice .....................................................................................................................8
3.3 HEALTH FACILITY GROUP ....................................................................................................................................8
3.3.1 Presentation: Measuring Maternal Mortality: The Potential of Health Facility Data.........................8 3.3.2 Presentation: The WHO Classification of during Pregnancy, Childbirth, and the Puerperium............9
3.3.3 Presentation: Implementing maternal death and near miss reviews to make pregnancy safer ........9
3.3.4 Maternal Mortality Audit: Botswana Experience..............................................................................10 3.3.5 Discussions and Practice ...................................................................................................................10
4 WORKSHOP SUMMARY ........................................................................................................................... 10
THE HOTEL INTERCONTINENTAL, NAIROBI, KENYA, 7-10 DECEMBER, 2010 ................................ 12
INTERCONTINENTAL NAIROBI, KENYA........................................................................................ 18
7-10 DECEMBER 2010 ........................................................................................................................... 18
7-8 DECEMBER 2010 (ANGLOPHONE)........................................................................................... 18
Annexes ANNEX 1. LIST OF PARTICIPANTS...............................................................................................................................12 ANNEX 2. AGENDA ....................................................................................................................................................18
iii
Acknowledgements
The workshop was coordinated by Eduard Bos, Emi Suzuki, Samuel Mills, and Bahie Rassekh
of the World Bank, Holly Newby and Liliana Carvajal of UNICEF, Ralph Hakkert of UNFPA,
and Lale Say and Doris Chou of WHO. Special thanks to Real Useful Travel and Marketing
Company for organizing the meeting in Nairobi, Kenya. Many thanks to the resource persons
listed in Annex 1 for facilitating the workshop. The contributions and participation of the
country experts listed in Annex 1 are much appreciated. Bahie Rassekh and Emi Suzuki
prepared the initial draft of this report, along with input from the UNICEF, WHO, and UNFPA
workshop coordinators and the workshop facilitators. The report was edited and translated into
French by the General Services Translation and Interpretation Unit, The World Bank
Contact persons:
Emi Suzuki, Human Development Network, World Bank. E-mail: [email protected]
Bahie Mary Rassekh, Human Development Network, World Bank. E-mail:
iv
Acronyms and Abbreviations
AIDS Acquired Immunodeficiency Syndrome
AMDF AIDS adjusted PMDF
DHS Demographic and Health Survey
GFR General Fertility Rate
HIS Health Information Systems
HIV Human Immunodeficiency Virus
ICD-10 International Statistical Classification of Diseases and Related Health Problems
(10th Revision)
ICPD International Conference on Population and Development
IHME Institute for Health Metrics and Evaluation
LTR Lifetime Risk
MDG Millennium Development Goal
MMR Maternal Mortality Ratio
MMRate Maternal Mortality Rate
PMDF Proportion of Maternal among female deaths 15-49
PRMRatio Pregnancy-Related Mortality Ratio
RAMOS Reproductive Age Mortality
RAPID Rapid Ascertainment Process for Institutional Deaths
TAG Technical Advisory Group
UC Berkeley University of California, Berkeley
UN United Nations
UNFPA United Nations Population Fund
UNICEF United Nations Children’s Fund
VR Vital Registration
WHO World Health Organization
v
1
INTRODUCTION
Improving sexual and reproductive health, which includes reducing maternal mortality, has
been a key concern of several international summits and conferences since the late 1980s,
including the International Conference on Population and Development (ICPD) in 1994 in
Cairo and the subsequent Fourth World Women’s Conference in 1995 in Beijing, which firmly
placed sexual and reproductive health and women’s health at the centre of social and economic
development. Their importance was again underlined at the Millennium Summit in 2000 when
one of the eight Millennium Development Goals (MDG) adopted was improving maternal
health (MDG5). Within the MDG monitoring framework (including 2005 World Summit
recommendations), the targets for improving maternal health are providing universal access to
sexual and reproductive health and reducing the maternal mortality ratio (MMR) by three-
fourths between 1990 and 2015.
However, assessment of the extent of the progress of MDG5 has been a challenge because of
lack of reliable data on sexual and reproductive health, and maternal mortality in particular.
This is of particular concern in countries in sub-Saharan Africa where maternal mortality is
high and the routine health information systems (HIS) are weak. Since the 1990s, the WHO,
UNICEF, UNFPA, and recently the World Bank have jointly developed five-yearly global
maternal mortality estimates using an approach that reconciles data from different sources, such
as facility-based data and population-based studies such as Demographic Health Surveys
(DHS), in order to obtain valid and internationally comparable country estimates. The approach
also involves the prediction of maternal mortality levels by using a statistical model for
countries where no nationally representative data drawn from standard methodologies exist.
The methodological challenges faced when estimating country-specific MMR point to a clear
need for generation of reliable and valid data to facilitate future estimates of sexual and
reproductive health indicators, particularly MMR in sub-Saharan where the existing data
sources are deficient. Accurate estimates of MMR require (i) complete records of all deaths in a
population, (ii) accurate attribution of the causes of death, and (iii) knowledge of the pregnancy
status of deceased women of reproductive age. However, none of the countries in sub-Saharan
Africa meet these criteria; the MMR for most countries in sub-Saharan Africa were either
derived from statistical models or entailed adjustment of the direct sisterhood estimates arising
from the DHS.
An estimated 358 000 maternal deaths occurred worldwide in 2008, a 34% decline from the
levels of 1990. Despite this decline, developing countries continued to account for 99% (355
000) of the deaths. Sub-Saharan Africa, alone accounted for 57% (204 000) of global maternal
deaths.
Overall, it was estimated that there were 42 000 deaths due to HIV/AIDS among pregnant
women in 2008. About half of those were assumed to be maternal. The contribution of
HIV/AIDS was highest in sub-Saharan Africa where 9% of all maternal deaths were due to
HIV/AIDS. Without these deaths, the MMR for sub-Saharan Africa would have been 580
maternal deaths per 100 000 live births instead of 640.
The MMR in 2008 was highest in developing regions (290) in stark contrast to developed
2
regions (14) and countries of the Commonwealth of Independent States (40). Among
developing regions, sub-Saharan Africa had the highest MMR at 640 maternal deaths per 100
000 live births in 2008. And three Sub Saharan countries, Chad, Guinea-Bissau, and Somalia,
have extremely high MMR (MMR ≥1000). Globally, the adult lifetime risk of maternal death
(the probability that a 15-year-old female will die eventually from a maternal cause) as
measured in 2008 is highest in sub-Saharan Africa (at 1 in 31).
The workshop, which took place in Nairobi, Kenya, was a joint World Bank, WHO, UNICEF
and UNFPA activity. It was composed of two 2-day meetings (back-to-back), separately
organized for 54 participants from 19 countries for the English session (December 7-8, 2010)
and 32 participants from 16 countries for the French session (December 9-10, 2010) in Africa.
The countries that participated in the English session were Botswana, Eritrea, Ethiopia, Ghana,
Kenya, Lesotho, Liberia, Malawi, Mozambique, Namibia, Nigeria, Somalia, South Africa,
Sudan, Swaziland, Tanzania, Uganda, Zambia, and Zimbabwe. The countries that participated
in the French session were Angola, Benin, Burkina Faso, Burundi, Cameroon, Central African
Republic, Chad, Côte d'Ivoire, Gabon, Guinea, Madagascar, Mali, Mauritania, Niger, São
Tomé and Principe, and Senegal. This report combines the proceedings of both the Anglophone
and Francophone sessions.
1.1 OBJECTIVES AND EXPECTED OUTCOMES
The objective of the workshop was to strengthen the capacity of national statisticians in each of
the participating countries on the approaches for improving data quality for estimating maternal
mortality.
The specific objectives and expected outcomes were as follows:
1.1.1 Specific objectives
• To establish a common understanding between UN agencies and national counterparts
on the approaches used for the recently developed 2008 maternal mortality estimates;
• To strengthen capacity of country statisticians in collecting, analyzing and use of
maternal mortality data;
• To identify needs and gaps in the generation, analysis, and use of data for estimating
maternal mortality.
1.1.2 Expected outcomes
• Common understanding between UN agencies and national counterparts on the
approaches used for the 2008 maternal mortality estimates;
• Improved technical expertise at country level in using methodologies and approaches to
measure maternal mortality.
1.2 OPENING SESSION
The workshop was officially opened by representatives from the four organizing agencies,
Khama Rogo for the World Bank, Richmond Tiemoko for UNFPA, Bo Pederson for UNICEF,
3
and Doris Chou for WHO, followed by opening remarks by Eduard Bos from the World Bank.
Dr. Bos conveyed the importance of the subject of improving reproductive and maternal health,
which has been the key concern of several international summits and conferences since the late
1980s. He defined MDG 5 and explained that WHO, UNICEF, UNFPA, and the World Bank
collaborated on a round of country-level estimates of maternal deaths. He provided some
background on the workshop, presented the workshop’s overall and specific objectives,
expected outcomes, and agenda.
2 PRESENTATIONS BY TECHNICAL EXPERTS
On the first day of the workshop, international experts made several presentations on the
approaches and methods for measuring maternal mortality. Annex 2 presents the agenda of the
meeting for the Anglophone and Francophone sessions. A summary of the presentations is
below.
2.1 SEXUAL AND REPRODUCTIVE HEALTH AND THE MDGS: WHO/UNICEF/UNFPA/WORLD BANK (H4) MATERNAL MORTALITY
ESTIMATES – OVERVIEW, HISTORY
Dr. Doris Chou presented an overview and history of the WHO/UNICEF/UNFPA/World Bank
maternal mortality estimates.
Dr. Chou described the health MDGs 4, 5, and 6, including indicators used to monitor progress.
She explained that there have been five-yearly estimates, separately for 1990, 1995, 2000, and
2005, that were based on an approach that encompassed different sources of data. There was a
decision to make more frequent updates, increasing the number of studies available from
countries. There was a review and revision of the previous approach and model. The
methodology changed, modeling maternal mortality over space and time. There was also a
hierarchical / multilevel linear regression model. A maternal mortality estimation interagency
group (MMEIG) was put in place.
Dr. Chou explained that a critical aspect of the Inter-Agency estimates of maternal mortality
was the technical collaboration with experts from academic institutions including University of
California, Berkeley, Harvard University, Johns Hopkins University, University of Texas,
Aberdeen, Umea University, and Statistics Norway. Another critical aspect was the country
consultation element. During this period of interaction with countries the estimates and
modeling methods are shared. Additional data are obtained and data quality reviewed, the
estimates are improved. The consultation improves the estimation process and builds a shared
understanding of country level activities in monitoring maternal mortality and the process
undertaken to derive internationally comparable estimates to enhance country use of the
estimates. This is a preliminary step in which countries have the opportunity to provide inputs
and share national data. Often it is found that the inter-agency best estimate and the country
reported figures may differ. In one country example, the Minister of Health said that the UN
estimate was much too high and that it was wrong. WHO sent a team of experts and an in-depth
4
study was conducted to review the data, uncover new data, analyze and discuss the results with
government and academic in the country. The result was an improved estimate.
Dr. Chou explained that following the consultation, continuous interaction with countries is
envisaged to strengthen capacity in data collection, reviewing data quality, updating the
database, supporting the use of data for decision making. Regional workshops are organized to
discuss these issues and the methodology of model based estimates. The technical
collaboration continues, such as during the upcoming TAG meeting in January 2011 during
which comments received during country consultation are reviewed for action. Lastly, the Inter
Agency maintains transparency in its work by making the database and the model publically
available
(www.who.int/reproductivehealth/publications/monitoring/9789241500265/en/index.html).
2.1.1 Discussion and observations
Following the introduction and background presentation, several participants raised concern on
the maternal mortality estimates for their individual countries. It was mutually agreed to move
forward with discussions on the methodology to address these concerns.
2.2 METHODS FOR MEASURING MATERNAL MORTALITY
Professor Kenneth Hill made a presentation regarding methods for measuring maternal
mortality. He defined a “maternal death” as the death of a woman while pregnant (or within 42
days of termination of pregnancy), irrespective of the duration and the site of the pregnancy,
from any cause related to or aggravated by the pregnancy or its management but not from
accidental causes (source: WHO 1993, 10th
revision of the ICD). He explained the difference
between a maternal death and a pregnancy-related death, where both have a temporal
relationship to the pregnant state, but a maternal death also has a causal relationship to the
pregnant state. The data collection method determines whether one measures maternal or
pregnancy-related deaths, since identifying maternal deaths requires either death certification
by an attending physician or a verbal autopsy. Household survey methods frequently used in
low/middle income countries simply ask time of death relative to pregnancy and thus measure
pregnancy-related death.
Dr. Hill then defined and provided characteristics of the MMR and explained that it is designed
to express direct or indirect obstetric risk, as opposed to the maternal mortality rate (MMRate),
which is a cause-specific death rate. He defined the Lifetime Risk (LTR) and Proportion
Maternal (PMDF) and how to calculate them.
Dr. Hill then explained the sources of data for measuring maternal mortality, including vital
registration, reproductive age mortality surveys (RAMOS), large population-based surveys,
national population censuses, facility-based studies, and statistical models. For large
population-based surveys, three survey methods of data collection/estimation were described:
The original sisterhood method, sibling history-based method, and identification of all female
deaths in the household in some reference period, which can be used in a census or large
5
survey. General problems with MMR measurement were explained, including that they are rare
events, that certain types of maternal death are hard to identify (especially abortion-related),
and that nonvital registration (VR) methods generally measure pregnancy-related mortality
ratio (PRMRatio).
He concluded by summarizing that maternal mortality is difficult to measure accurately, even
in countries with a complete VR, that in countries lacking complete VR, no approach is
guaranteed to give accurate estimates, and finally, that in the long run, it is essential to improve
VR.
2.3 GLOBAL MATERNAL MORTALITY ESTIMATES AND TRENDS, 1990-2008
Dr. John Wilmoth’s presentation had three parts: 1. Data and methods used for creating the new
UN estimates; 2. Key findings regarding levels and trends of maternal mortality (especially for
Sub-Saharan Africa (SSA); and 3. Comparison to other estimates (including national sources).
Under the first topic, Data and Methods for Estimating Maternal Mortality, several areas were
discussed. These include coverage of estimates, sources of maternal mortality data used for the
2008 estimates, data types worldwide and in SSA, definitions of maternal death and pregnancy-
related death, PMDF, Mortality from all causes, envelope adjustment, data on live births, data
adjustments, underreporting of maternal or pregnancy-related deaths, and multilevel regression
models. With regards to the regression model, the model was specified, including dependent
variables, covariates, and offset terms. Three variants of the model were explored, using
PMDFna
, AMDFna
, and MMRna
as the dependent variable. It was explained that for each
covariate or offset variable, complete annual data series were either obtained or created. For the
regression model, average values of covariates and offset variables were computed over time
intervals matched to each PMDF observation. The final model included random effects for both
countries and regions to provide a simple means of depicting those components of the
variability in PMDFna
that are not well described by the simple model with three covariates.
He also presented the key findings of the 2008 MMR estimates and explained why the
estimates may differ from national estimates.
2.3.1 Discussion and observations
Participants suggested the addition of other maternal health variables to the model, for example
indicators related to emergency care. Some participants also expressed interest in having
information on national policies reflected on the model. Dr. Wilmoth explained that the lack of
cross country data on additional variables or national policies would make it difficult for them
to be included.
6
2.4 DEMONSTRATION OF PROPOSED MATERNAL MORTALITY ESTIMATION
DASHBOARD
Holly Newby proposed the idea of the maternal mortality estimation dashboard and provided a
demonstration for participants. As an example, she showed several graphs with the 2008
estimates and trends, with the information on inputs used, including Gross Domestic Product,
skilled birth attendance and general fertility rate(GFR).
2.4.1 Discussions and observations
• In general, participants and facilitators thought having the dashboard was a good idea
because it was informative for countries for both inputs and outputs.
• Although the prototype did not include country-generated MMR data (directly obtained
from surveys, etc.), the final version of the dashboard will include both the MMR
estimates from the country as well as the interagency estimates.
• A suggestion was made for the dashboard to include other indicators related to maternal
health to provide a larger context for interpreting trends. Although the initial version of
the dashboard will not include data beyond maternal mortality, the model input
variables of GDP, skilled birth attendance and GFR will be included. This idea will be
explored for inclusion in a subsequent version of the dashboard.
• A number of participants noted that because the maternal mortality estimation
dashboard is based on the DevInfo platform, it would be ideal to be able to link the
dashboard with their own country’s DevInfo application. This idea can be explored,
although it is important to note that even if an individual country has new input data, the
model is based on all available data globally and thus cannot be rerun at the country
level using a single country’s input data.
3 GROUP SESSIONS AND THE WAY FORWARD
There were three groups in the group sessions, with a focus on the following themes: census,
household surveys, and routine health data.
3.1 HOUSEHOLD SURVEY GROUP
3.1.1 Presentation: Estimating Maternal or Pregnancy-Related Mortality Using Household Surveys
In this session, Dr. John Wilmoth discussed the three survey methods of data collection,
including the original sisterhood methods (only listed, not discussed), the sibling history-based
method, and the identification of all female deaths in the household in some reference period.
This presentation focused on the full sibling history, including background information of this
7
method, data collected, and examples. The presentation then introduced data evaluation and
adjustment, including definitions and acronyms, information regarding sampling errors, data
quality indicators, recall bias, selection bias, the Gakidou-King Estimation Method and
adjustments. After this, data evaluation was introduced. This included potential data
evaluations, mortality level, Coale-Demeny Model Life Tables, sex differentials in adult
mortality, age patterns of adult mortality, and proportions of pregnancy-related deaths by age
group. The presentation ended with a section on computations. The following steps were
outlined: Calculate pregnancy-related mortality rates, age-standardize overall pregnancy-related
mortality rate, standardize the total pregnancy-related death rate to the household female
population, use the general fertility rate to calculate the pregnancy-related mortality ration per
100,000 live births, calculate pregnancy-related mortality ration using GFR, age-standardize
overall proportion of deaths pregnancy-related 15-49 years, calculate age-standardized
proportion of pregnancy-related deaths of women 15-49 years, and the last step, which was to
estimate the risk that a woman surviving to age 15 will die a pregnancy-related death. There
was an explanation of the LTR equation, and the session concluded with practice calculations
completed by participants.
3.1.2 Discussions and practice
After Dr. Wilmoth’s presentation, participants worked on an exercise which involved
identifying the right data from DHS reports on-line, or in copies of the report provided by
participants. They entered the data in worksheets and calculated the various maternal mortality
indicators. Participants were guided through the process of calculation and interpretation of the
results. A volunteer from the group was selected to present conclusions from the household
survey analysis to all the participants during the subsequent plenary session.
3.2 CENSUS GROUP
3.2.1 Presentation: Estimating Pregnancy-Related Mortality from the Census
Dr. Kenneth Hill’s presentation focused on ways of estimating pregnancy-related mortality
from the census. There are three components of PRMRatio: 1. Deaths of women of
reproductive age (D); 2. The proportion of those deaths that were pregnancy-related (PPR); and
3. Births (B). The formula for the PRMRatio is (D*PPR*100,000)/B. Therefore the evaluation
focuses on D, PPR and B.
Dr. Hill used the example of South Africa to illustrate ways to use data from the various
censuses to evaluate the numbers of deaths of women of reproductive age. He explained that
this involves evaluating female deaths at all ages post-childhood. It is most important that
deaths of older women are recorded. Numbers of deaths are evaluated by comparison with the
population age distribution. He explained the key assumptions: The methods assume that the
errors of reporting (deaths and population) are distributed proportionately by age, that the
population is closed to migration, the Brass Growth Balance method assumes that the
population is demographically stable, and the General Growth Balance method replaces the
assumption of stability by using data from two censuses, but assumes that the age pattern of
8
deaths in the intercensal interval is approximated by the observed pattern. The General Growth
Balance Method was then explained in greater detail and illustrated using Honduras as an
example. Dr. Hill then discussed the calculations, including the entry rate, the growth rate, the
observed death rate and the computation of adjustment factor for deaths. He went over
spreadsheets, graphs, and interpretations.
Dr. Hill then discussed evaluating the proportion of deaths that were pregnancy-related, and
evaluating the numbers of births. He explained that the evaluation is of recent fertility against
lifetime fertility (P/F Ratios), and then explained the P/F Ratios in further detail, including the
principle, application, complications. He explained the interpretation of the data, and concluded
with the Pregnancy-Related Mortality Ratio (PRMRatio).
3.2.2 Discussions and Practice
After Dr. Hill’s presentation, participants spent the remaining time practicing the methods by
entering data into the excel file provided by Dr. Hill. Dr. Hill guided the participants so they
could visually examine the results in the graph. Dr. Hill answered questions and accompanied
participants as they interpreted the results. Dr. Hill and the participants prepared the
presentations on key findings and recommendations from the group.
3.3 HEALTH FACILITY GROUP
3.3.1 Presentation: Measuring Maternal Mortality: The Potential of Health Facility Data
Dr. Carla Abou-Zahr’s presentation included the following topics: principles for using facility
data; advantages and limitations of facility data; data requirements; data quality assessment and
adjustment; methods to improve data completeness and quality; technical resources; and
country examples.
She began by defining maternal death and pregnancy-related death, and explained that the data
sources that identify “true” maternal deaths require medical determination of cause. She
discussed health facility records and health management information systems (HMIS) as well
as strengths and limitations of facility-based HMIS data. She discussed how we can further tap
the potential of facility data and some of the reasons for the weakness of routine HMIS
reporting. She introduced a tool called RAPID (Rapid Ascertainment Process for Institutional
Deaths) to review hospital records for all deaths of women aged 15 to 49 years in order to
improve identification of all maternal deaths in facilities. Dr. Abou-Zahr then discussed
identifying maternal deaths in communities and options for identifying home deaths,
reconciling facility data and community reporting (capture-recapture), and conditions for using
routine HMIS reports from health facilities for monitoring. She discussed country strategies for
improving maternal mortality data, technical resources, and major categories of MMR data
sources.
9
3.3.2 Presentation: The WHO Classification of during Pregnancy, Childbirth, and the Puerperium
Dr. Doris Chou presented work on the WHO Classification of during Pregnancy, Childbirth,
and the Puerperium. She explained that inconsistency in death attribution leads to
misclassification of maternal deaths extracted from vital registration, which in turn may bias
the understanding of the magnitude and causes of maternal death. The WHO Classification of
maternal deaths is intended to simplify and standardize the capture of maternal deaths from VR
and other sources of data. The current version is based upon ICD 10 codes. She explained
principles of the classification system, defined maternal death in relation to the ICD 10 and
explained the overall structure of classification. Dr. Chou defined and explained the
relationship between contributing conditions and underlying cause and immediate causes of
death, and provided examples. She closed her presentation with conclusions and one
recommendation, that once the maternal death classification system is published, it should be
adopted by all countries.
3.3.3 Presentation: Implementing maternal death and near miss reviews to make pregnancy safer
Professor Gwyneth Lewis had six main parts to her presentation: 1. Underlying principles; 2.
The Beyond the Numbers approaches; 3. Which deaths? Which approach?; 4. Steps in the
process; 5. Other issues to consider; 6. Detailed descriptions of community based death
reviews, facility based deaths reviews and near miss reviews.
She explained that a large percentage of maternal deaths occur due to substandard care and
preventable causes. The MMR is a crude estimate, does not give causes of death, does not
address intra-country variations, and is often out of date. It often appears overwhelming to
policy makers or health care planners and does not have insight into avoidable or remediable
factors. She stressed that we need to look beyond the numbers at why mothers really die.
There are three types of delays/barriers to care: socio-cultural (status of women, traditional
beliefs, practices); lack of access to care (availability, transport); and poor quality care, poor
resources. Professor Lewis talked about the five approaches to look “Beyond the Numbers” and
shared characteristics between all these approaches. She discussed the maternal mortality
surveillance cycle, as well as investigations into causes of deaths such as verbal autopsy,
facility based death reviews, learning from near miss case reviews, and Confidential Enquiries
into Maternal Deaths, and factors in deciding which approach to adopt. She discussed types of
maternal deaths and methods to identify and review maternal deaths. She went through steps in
the process, including setting up an overview committee at the start, collecting data including
issues of confidentiality, legal and ethical considerations, analyzing the results both
quantitatively and qualitatively, and disseminating findings. Professor Lewis then discussed
deaths in the community, community-based reviews, and community data collection. She
discussed facility based reviews, types, scope, and potential difficulties and gains. She
discussed “near-miss” cases, why to review them, uses of these data, approaches to reviewing
these cases, including card based surveillance systems. Professor Lewis closed by giving some
10
reference sites: www.cmace.org.uk, www.nice.org.uk, [email protected],
www.dh.gov.uk, www.npsa.org.uk, www.rcog.org.uk, www.cqc.org.uk
3.3.4 Maternal Mortality Audit: Botswana Experience
Boitumelo Thipe, the National Safe Motherhood Program coordinator, Botswana, made this
presentation regarding the Botswana experience with Maternal Mortality Audits. She described
the health system of Botswana and some of the history of the maternal death review process.
She explained that for purposes of counter-checking, maternal mortality is captured by two
parallel systems, one based in SRH and the other in HSU, and then described both processes
and the SRH/HSU system linkages. She outlined the maternal death notification process for
audit purposes, the composition of the maternal mortality committees, and described the 2008
national maternal mortality report. Finally, she discussed the maternal deaths by age, by direct
and indirect causes, the proportion of pregnancy complications, the MMR and MDG trend over
the past five years, and contributory factors per cause of death. She concluded by discussing
the improvement of clinical care using maternal death audits, some challenges, and areas that
need strengthening.
3.3.5 Discussions and Practice
The participant from Madagascar shared some approaches taken at country level to standardize
and improve cause of death attribution. This results in improved cause of death attribution.
Further improvements is anticipated with an international standard on maternal cause of death
attribution, such as proposed in the WHO Classification of Deaths in Pregnancy, Childbirth and
the Puerperium.
4 WORKSHOP SUMMARY
Holly Newby from UNICEF presented a recap of the experiences in the working groups and
draft recommendations. Her summary included the following points:
1. General technical issues
a. Definitions are important
i. MMR versus MMRate
ii. Maternal deaths versus pregnancy-related deaths
b. The MMR gives a false sense of precision
c. Maternal mortality data comes from a variety of sources
i. Each source has advantages and limitations
ii. Even vital registration, misreporting well-documented
iii. All maternal mortality data require careful, critical evaluation
2. UN interagency estimates
11
a. UN interagency estimates address data limitations and provide comparable trend
estimates for 172 countries
b. To produce the UN interagency estimates, a series of decisions had to be made
i. Use (adequate) civil registration data directly for trend estimation
ii. Model PMDF (not MMR) for all other countries
iii. Choice of covariates
iv. How to handle HIV
v. Envelope of all cause female deaths; births
c. UN estimates: Special efforts made to:
i. Document all decisions
ii. Make all data and calculations publically available
3. Working groups
a. Household survey group (Direct sisterhood method)
i. Important source of data for sub-Saharan Africa
ii. Mortality underestimated overall
iii. Confidence intervals very wide (difficult to interpret trends)
iv. Recommendations—
1. Improve data collection
2. Necessary to review data quality, perform critical checks
b. Census group
i. A number of sub-Saharan African countries have (or will) include
questions on deaths in household (and whether pregnancy related)
ii. Recommendations –
1. Data collection should be standardized
2. Need to evaluate all data carefully
3. Even if they appear implausible, keep original data because may
be possible to adjust
4. Need mechanisms for continued exchange and technical
assistance
c. Health facility group
i. Strength -- Provides information on cause of death; produced regularly
and all over country
ii. Weakness -- Facility data often incomplete; problems with
denominators; bias because only based on facility deliveries
iii. How to strengthen
1. Need to regularly review and evaluate facility data
2. Health care providers: Need to increase skills and understanding
3. Audits important – help improve data and take you beyond
numbers to underlying causes and circumstances
4. Main concluding points:
a. Critical data reviews are essential
b. Look across all potential data sources
c. Important to look beyond the MMRatio
12
Annex 1. List of participants
THE WORLD BANK
Improving national capacity to track maternal mortality towards the attainment of the MDGs
The Hotel Intercontinental, Nairobi, Kenya, 7-10 December, 2010
English Session Participants (December 7-8, 2010)
Botswana Boitumelo Thipe, Sexual Reproductive Health Division, Department Public Health, Ministry of Health
Botswana Babuang Tlhomelang, Central Bureau of Statistics
Botswana Kelebetse Mbiganyi, Central Bureau of Statistics
Botswana Lucy Sejo Maribe, Nurse, DPM/FRH, WHO Botswana
Botswana Peter Beat Gross, Social Policy Specialist, UNICEF
Eritrea Mismay Ghebrehiwet, Advisor to Minister of Health (MOH)
Eritrea Yodit Hiruy, UNICEF
Ethiopia Gebeyehu Horjo, Central Bureau of Statistics
Ethiopia Sahelu Gelaye, Central Bureau of Statistics
Ghana Philomena Efua Nyarko, Deputy Government Statistician, Statistical Service
Kenya Andrew Imbwaga, Kenya National Bureau of Statistics
Kenya Henry Osoro, Kenya National Bureau of Statistics
Kenya Patricia Elung'ata, DSS Data Manager, African Population and Health Research Center
Lesotho Nonkosi Tlale, UNFPA
13
Lesotho Thabelo Ramatlapeng, UNFPA
Liberia Woseh Gobeh, Reproductive Health Commodity Officer at the MOH
Malawi George Chapotera, Principal Economist, MOH Central M&E Division
Malawi Mylen Mahowe, Demography and Social Statistics, NSO
Mozambique Carlos Arnaldo, Centro de Estudos Africanos (Centre for African Studies)
Mozambique Cassiano Soda Chipembe, Instituto Nacional de Estatistia (National Institute of Statistics)
Namibia Mutambani Mwakamui, Ministry of Health
Namibia Ndapandula Ndiknetepo, Central Bureau of Statistics
Nigeria Agosa Olusegun, Statistician, Dept of Health Planning, Research and Stats, Ministry of Health
Nigeria Denis Jobin, Chief Planning, M&E Coordinator, UNICEF
Nigeria Godwin A. Iro, National Bureau of Statistics
Nigeria Ossideko Olusola, National Population Commission
Nigeria Utibe.Abasi Essien Urua, National Primary Health Care Development Agency
Somalia Ahmed D. Farah, Ministry of Planning
Somalia Rogaia Abuelgasim, UNFPA
Somalia Stephen Macharia, UNFPA
South Africa Bjorn Gelders, M&E Officer, UNICEF South Africa
South Africa Khangelani Zuma, Human Sciences Research Council
South Africa Maletela Tuoane-Nkhazi, Health and Vital Statistics Division, Statistics South Africa
South Africa Nat Khaole, National Department of Health
South Africa Mosidi Sarah Nhlapo, Demographer,
South Africa Sean Jooste, Human Sciences Research
14
Statistics South Africa, Census Analysis
Council
Sudan Barbara Akita Kibos, MOH- Government of South Sudan
Sudan Majak Makal Adhil, Southern Sudan Centre for Census Statistics and Evaluation
Sudan Richard Lino Laku, Director, Monitoring and Evaluation, MOH-GOSS (Govt of Southern Sudan)
Swaziland Dlamini Nombulelo, Statistician, Central Bureau of Statistics
Swaziland Duduzile Dlamini, National Population Unit
Swaziland Phumzile Mabuza, SRHU, MNCH Manager
Swaziland Simelane Zanela, HMIS Coordinator, Ministry of Health
Tanzania Aldegunda Komba, Statistician / Demographer, National Bureau of Statistics
Tanzania Asia Hussein, UNICEF
Tanzania Edith Mbatia, UNICEF
Tanzania Mayasa Mwinyi, Office of Chief Gov. Statistics
Uganda Anthony Mbonye, Commissioner, Health Services, Ministry of Health
Uganda Mark Kajubi, Sr Statistician, Uganda Bureau of Statistics
Uganda Miriam Sentongo, Reproductive Health Division, Ministry of Health
Uganda Vincent Ssenono, Uganda Bureau of Statistics
Zambia Sheila Mudenda, Central Bureau of Statistics
Zimbabwe Godfrey Matsinde, ZIMSTAT
Zimbabwe Winston Chirombe, Ministry of Health and Child Welfare
French Session Participants (December 9-10, 2010)
Angola Helga de Freitas, Médica de Saúde Pública, Revitalização do Sistema Municipal de Saúde
Benin Ahovey A Elise C, Ingeneur demographe, Institut National de la Statistique et de l'Analyse Economieque / Ministry of
15
Development
Benin Hyacinthe Ahomlanto, Médecin Gynécologue- Obstétricien, Direction de la Santé de la Mère et de l’Enfant/service Santé Maternelle et Infantile
Burkina Faso Doamba Jean Edouard, Direction generale de l'information et des statistiques sanitaires
Burkina Faso Tingueri Rose Koirine, Ingenieure Demographe-Geographe
Burundi Deogratias Buzingo (Chief Section of Demographic studies and Statistics at the National Institute of Statistics)
Burundi Noe Nduwabike, Chef Division Traitement et Analyse, RGPH
Burundi Dionis Nizigiyimana, EPISTA, MoH
Cameroun KAMGHO TEZANOU Bruno Magloire, Institut National de la Statistique
Cameroun Ndong Ngoe Constant, Epidemiologist from Ministry of Public Health
Côte d'Ivoire Yao Koffi Edmond, Chief of Demographic and Social Statistics at the National Statistics Institute
Côte d'Ivoire Youan Rodolphe Tian Bi, Assistant Suivi-Evaluation, UNFPA 01 BP 1747, Abidjan 01
Gabon Yolande Vierin, Ministère de la santé
Guinée Diallo M.D. Dile, Ministère du plan
Madagascar Eugene Kongnyuy, CTA Maternal Health
Madagascar Randretsa Iarivony, Director of Demography and Social Statistics from the National Institute of Statistics in Madagascar
Mali Mamadou Diop, Chef Unite Statistique, CPS/Secteur Sante
Mauritanie Brahim Ould Mohamed ould Amar, Chef de Service des Methodes et du Suivi des Indicateurs (MOH), focal point
Niger Abduol Rachid Fatima, Departement centrale de la sante maternelle et infantile,
Niger Argoze Moussa, Direction des Statistiques et des Etudes Demographiques et Sociales
16
MOH
(DSEDS)
Niger Haladou Moussa, Technicien Supérieur en Santé Publique, DSS/RE-MSP, Div. Formation et Recherche
Niger Mariama Djakounda Pascal, UNFPA
Niger Sani Oumaro, Direction des Statistiques et des Etudes Demographiques et Sociales (DSEDS)
Niger Yaroh Asma Galo, Departement centrale de la sante maternelle et infantile, MOH
République Centre-Africaine Aguide Soumouk, Responsable suivi évaluation, DSFP, Ministère Santé RCA
République Centre-Africaine Ali Blaise Bienvenu, StatALI Blaise Bienvenu, Statisticien-Démographe, Chef de Service des Statistiques Démographiques
São Tomé and Principe Agostinho Miguel Soares Batista de Sousa, Ministère de la santé
Sénégal Aida Tall, Médecin, en service à la DSR du Ministère de la Santé et de la Prévention
Tchad Fatime Marthe Koulassengar, UNICEF
Tchad Gnayam An Koumtingue, Chargee du Suivi et Evaluation, UNICEF N'Djamena, Unite Suivi et Evaluation
Tchad Lam nee Mai Service, UNICEF
Tchad Prosper Lawe Ngaindandji, Demographe, INSEED
Resource Persons
Carla Abou-Zahr Consultant, WHO 6 chemin des Fins, le Grand Saconnex, 1218 Geneva, Switzerland
Kenneth Hill Harvard University SPH, Harvard Init on Global Health Third Floor 104 Mt Auburn St Cambridge MA 02138
Gwyneth Lewis National Clinical Director for Maternal Health and Maternity Services, Department of Health, England 202 Wellington House, 133-155 Waterloo
John Wilmoth Associate Professor in the Department of Demography, University of California, Berkeley Department of Demography
17
Road, London SE1 8UG
University of California, Berkeley 2232 Piedmont Avenue Berkeley, California 94720-2120
Holly Newby Division of Policy and Planning Strategic Information Section UNICEF Three United Nations Plaza, New York, NY 10017
Liliana Carvajal Division of Policy and Planning Strategic Information Section UNICEF Three United Nations Plaza, New York, NY 10017
Ralph Hakkert Technical Advisor Population and Development Branch 220 East 42nd Street 17st Floor, New York
Lale Say MD and Epideomiologist, Department of Reproductive Health and Research, World Health Organization HQ Avenue Appia, 1211 Geneva 27, Switzerland
Doris Chou Medical Officer, World Health Organization HQ, Geneva Avenue Appia, 1211 Geneva 27, Switzerland
Eduard Bos Lead Population Specialist (AFTHE) Washington D.C. The World Bank MSN J10-1004 1818 H Street NW Washington, DC 20433
Samuel Mills Health, Nutrition & Population (HDNHE) The World Bank MSN G7-701 1818 H Street NW Washington, DC 20433
Emi Suzuki Research Analyst (HDNHE) The World Bank MSN G7-701 1818 H St NW Washington, DC 20433
Khama Odera Rogo Lead Health Specialist, The World Bank (IFC) Nairobi The World Bank, Nairobi, Kenya (IFC) Mail Stop: NBOWB
Bahie Mary Rassekh Health Consultant (HDNHE) The World Bank MSN G7-701 1818 H St NW Washington, DC 20433
18
Annex 2. Agenda
THE WORLD BANK
Improving national capacity to track maternal mortality towards the attainment of the MDG5
InterContinental Nairobi, Kenya 7-10 December 2010 7-8 December 2010 (Anglophone) 9-10 December 2010 (Francophone)
Agenda
Day One
Time Description Presenter
08:30-09:00
Registration Susan Oloo-Oruya
09:00-09:30
Welcome/introductory remarks Khama Rogo, World Bank; Richmond Tiemoko, UNFPA Bo Pederson, UNICEF Doris Chou, WHO
09:30-
09:45
Background & agenda
Objectives/expected outcomes
Eduard Bos, World Bank
09:45-10:00
Sexual and reproductive health and the MDGs
Maternal mortality - overview, history, why the five-yearly inter-agency maternal mortality estimates
Discussion
Doris Chou, WHO
10:00-
11:15
Methods for Measuring Maternal Mortality -- Definitions -- Sources of data to estimate maternal mortality
-- Methods of data collection, estimation
Discussion
Kenneth Hill, Harvard University
11:15-11:30
Tea break
11:30-12:30
Global maternal mortality estimates & trends, 1990-2008
-- Data and methods used for creating the new UN estimates
-- Key findings regarding levels and trends of maternal mortality (especially for Sub-Saharan
John Wilmoth, UC Berkeley
19
Africa)
-- Comparison to other estimates (including national sources)
Discussion
12:30-13:45
Lunch
13:45- 14:30
Global maternal mortality estimates & trends, 1990-2008 (continued)
John Wilmoth, UC Berkeley
14:30-14:45
Demonstration of Proposed Maternal Mortality Estimation Dashboard
Holly Newby, UNICEF
14:45-15:00
Group Photo / Tea break
15:00-17:30
Working groups - Measuring maternal mortality (3 groups)
1. Household survey group
-- Advantages and limitations -- Questionnaire/data required -- Data Evaluation and adjustment -- Resources available to developing countries -- Country examples
2. Census group
-- 2010 Principles and Recommendations for Population and Housing Censuses
-- Advantages and limitations
-- Questionnaire/data required
-- Data Evaluation and adjustment
-- Resources available to developing countries
-- Country examples
3. Health facility group
--Presentation of attributes of health facility data
--Maternal health audits
-- Country examples and application of health facility data and death audits (Botswana)
-- Death classification discussion
-- Discussion and preparation of recommendations
John Wilmoth, UC Berkeley
Holly Newby, UNICEF
Liliana Carvajal, UNICEF
Kenneth Hill, Harvard University Richmond Tiemoko, UNFPA
Emi Suzuki, World Bank
Bahie Rassekh, World Bank
Carla Abou-Zahr, WHO Consultant
Gwyneth Lewis, Dept of Health, UK
Doris Chou, WHO
Eduard Bos, World Bank
18:00 Cocktail
20
Day Two
Time Description Presenter
09:00-12:00
Working groups (continued)
1. Household survey group
2. Census group
3. Health facility group
John Wilmoth, UC Berkeley
Holly Newby, UNICEF
Liliana Carvajal, UNICEF
Kenneth Hill, Harvard University Richmond Tiemoko, UNFPA
Emi Suzuki, World Bank
Bahie Rassekh, World Bank
Carla Abou-Zahr, WHO Consultant
Gwyneth Lewis, Dept of Health, UK
Doris Chou, WHO
Eduard Bos, World Bank
12:00-13:30
Lunch
13:30-14:15
Working group presentations, discussions and recommendations (Household survey group)
Group rapporteurs
14:15-15:00
Working group presentations, discussions and recommendations (Census group)
Group rapporteurs
15:00-15:15
Tea Break
15:15-16:00
Working group presentations, discussions and recommendations (Health facility group)
Group rapporteurs
16:00-
17:20
Recap of the experiences in the working groups and draft recommendations
Discussion of the recommendations on how to improve maternal mortality estimation at country level
Wrap-up /way forward
Holly Newby, UNICEF
WHO
UNICEF
World Bank
UNFPA
17:20-17:30
Closing remarks 4 Agency representatives
Chairperson:
Tuesday, 07 December 2010 Morning session: Khama Rogo, World Bank Wednesday, 08 December 2010
21
Afternoon session: Carla Abou-Zahr, WHO Consultant