SDMX Metadata Concepts

Download PDF

0.a. Goal

Goal 3: Ensure healthy lives and promote well-being for all at all ages

0.b. Target

Target 3.1: By 2030, reduce the global maternal mortality ratio to less than 70 per 100,000 live births

0.c. Indicator

Indicator 3.1.1: Maternal mortality ratio

0.d. Series

Applies to all series

0.e. Metadata update

2023-03-31

0.g. International organisations(s) responsible for global monitoring

World Health Organization (WHO). Department of Sexual and Reproductive Health and Research.

1.a. Organisation

World Health Organization (WHO). Department of Sexual and Reproductive Health and Research.

2.a. Definition and concepts

Definition:

The maternal mortality ratio (MMR) is defined as the number of maternal deaths during a given time period per 100,000 live births during the same time period. It depicts the risk of maternal death relative to the number of live births and essentially captures the risk of death in a single pregnancy (proxied by a single live birth).

Concepts:

In the International statistical classification of diseases and related health problems (ICD) WHO defines the following:

Maternal death: The death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration and site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management (from direct or indirect obstetric death), but not from unintentional or incidental causes.

A death occurring during pregnancy, childbirth and puerperium (also known as a pregnancy-related death): The death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the cause of death.

2.b. Unit of measure

Maternal deaths per 100,000 live births

2.c. Classifications

Maternal deaths are classified according to the International statistical classification of diseases and related health problems (ICD) definition. The specific codes used under ICD-10 (the 10th revision of the ICD) to define a maternal death are: O00-O96; O98, O99 and A34.

ICD-11 (the 11th revision of the ICD) was adopted by the World Health Assembly in May 2019 and comes into effect on 1st January 2022. Further information is available at: www.who.int/classifications/icd/en/ The coding rules related to maternal mortality are being edited to fully match the new structure of ICD-11, but without changing the resulting statistics. The ICD-11 rules can be accessed in the reference guide of ICD-11, at https://icd.who.int . Forthcoming releases from 2022 onwards will transition to use ICD-11 coding. Care has been taken to ensure that the definition of maternal death used for international comparison of mortality statistics remains stable over time, but the word “unintentional” has been used in the ICD-11 definition in place of the word “accidental” which was previously used, in ICD-10.

3.a. Data sources

Please see Sections 3.1 and 3.2 of the report: Trends in maternal mortality 2000 to 2020: estimates by WHO, UNICEF, UNFPA, World Bank Group and UNDESA/Population Division. Geneva: World Health Organization; 2023.

3.b. Data collection method

The United Nations Maternal Mortality Estimation Inter-Agency Group (UN MMEIG) – comprising WHO, UNICEF, UNFPA, the World Bank Group and the United Nations Population Division (UNDESA/Population Division) maintains an input database consisting of maternal mortality data from civil registration, population-based surveys, surveillance systems, censuses, and other specialized studies/surveys. This database is updated before the release of every new round of estimates and is used to calculate the proportion of maternal deaths (PM) among women of reproductive age (WRA). The maternal mortality ratio (MMR) is then calculated as MMR = PM(D/B); where "D" is the number of all-cause deaths among women WRA and "B" is the number of live births. The number of live births is based upon the World Population Prospects published by UNDESA/Population Division.

Statistical modelling is undertaken to generate comparable country, regional, and global level estimates. Adjustments are made according to the data source type (See Section 4e below). The analysis accounts for stochastic errors, sampling error in the data source, errors during data collection and processing, and other random error. The model's fit is assessed by cross-validation.

3.c. Data collection calendar

The input datasets are updated prior to each new publication round of the maternal mortality ratio (MMR) estimates. Source data are collected by countries, typically annually for civil registration and vital statistics (CRVS) sources, every 3-5 years for specialized reviews, every 5-7 years for population-based surveys, and every 10 years for censuses.

3.d. Data release calendar

The maternal mortality estimates are updated approximately every 2-3 years.

3.e. Data providers

National-level data providers are typically statistical offices, specialized epidemiology monitoring authorities and/or Ministry of Health.

3.f. Data compilers

The United Nations Maternal Mortality Estimation Inter-Agency Group (UN MMEIG) – comprising WHO, UNICEF, UNFPA, the World Bank Group and the United Nations Population Division (UNDESA/Population Division) of the Department of Economic and Social Affairs.

3.g. Institutional mandate

World Health Organization (WHO) is the custodian UN agency for the maternal mortality ratio.

4.a. Rationale

All maternal mortality indicators include a point-estimate and an 80% uncertainty interval (UI). Both point-estimates and 80% UIs should be taken into account when assessing estimates.

For example: “The estimated 2020 global MMR is 223 (UI 202 to 255).”

This means:

  • The point-estimate is 223 and the 80% uncertainty interval ranges 202 to 255.
  • There is a 50% chance that the true 2020 global MMR lies above 223, and a 50% chance that the true value lies below 223.
  • There is an 80% chance that the true 2020 global MMR lies between 202 and 255.
  • There is still a 10% chance that the true 2020 global MMR lies above 255, and a 10% chance that the true value lies below 202.

Other accurate interpretations include:

  • We are 90% certain that the true 2020 global MMR is at least 202.
  • We are 90% certain that the true 2020 global MMR is 255 or less.

The amount of data available for estimating an indicator and the quality of that data determine the width of an indicator’s UI. As data availability and quality improve, the certainty increases that an indicator’s true value lies close to the point-estimate.

4.b. Comment and limitations

The extent of maternal mortality in a population is essentially the combination of two factors:

  1. The risk of death in a single pregnancy or a single live birth.
  2. The fertility level (i.e. the number of pregnancies or births that are experienced by women of reproductive age).

The maternal mortality ratio (MMR) is defined as the number of maternal deaths during a given time period per 100 000 live births during the same time period. It depicts the risk of maternal death relative to the number of live births and essentially captures (i) above.

By contrast, the maternal mortality rate (MMRate) is calculated as the number of maternal deaths divided by person-years lived by women of reproductive age. The MMRate captures both the risk of maternal death per pregnancy or per total birth (live birth or stillbirth), and the level of fertility in the population.

In addition to the MMR and the MMRate, it is possible to calculate the adult lifetime risk of maternal mortality for women in the population. An alternative measure of maternal mortality, the proportion of deaths among women of reproductive age that are due to maternal causes (PM), is calculated as the number of maternal deaths divided by the total deaths among women aged 15–49 years.

4.c. Method of computation

The maternal mortality ratio (MMR) can be calculated by dividing recorded (or estimated) maternal deaths by total recorded (or estimated) live births in the same period and multiplying by 100 000. Measurement requires information on pregnancy status, timing of death (during pregnancy, childbirth, or within 42 days of termination of pregnancy), and cause of death.

The MMR can be calculated directly from data collected through vital registration systems, household surveys or other sources. There are often data quality problems, particularly related to the underreporting and misclassification of maternal deaths. Therefore, data are often adjusted in order to take these data quality issues into account. Some countries undertake these adjustments or corrections as part of specialized/confidential enquiries or administrative efforts embedded within maternal mortality monitoring programmes.

Bayesian maternal mortality estimation model (the BMat model):

Estimation and projection of maternal mortality indicators are undertaken using the BMat model. This model is intended to ensure that the MMR estimation approach is consistent across all countries but remains flexible in that it is based on covariate-driven trends to inform estimates in countries or country-periods with limited information; captures observed trends in countries with longer time series of observations; and takes into account the differences in stochastic and sampling errors across observations.

The model is summarized as follows:

log ⁡ E P M N A = b 0 + b 1 log ⁡ G D P + b 2 log ⁡ G F R + b 3 S B A + γ j + φ k

Where:

E P M N A = the expected proportion of non-HIV-related deaths to women aged 15–49 years that are due to maternal causes [NA = non-HIV; formerly it referred to “non-AIDS”]

GDP = gross domestic product per capita (in 2011 PPP US dollars)

GFR = general fertility rate (live births per woman aged 15–49 years)

SBA = proportion of births attended by skilled health personnel

γ j = random intercept term for country j

φk = random intercept term for region k.

For countries with data available on maternal mortality, the expected proportion of non-HIV-related maternal deaths was based on country and regional random effects, whereas for countries with no data available, predictions were derived using regional random effects only.

The resulting estimates of the E P M N A were used to obtain the expected non-HIV MMR through the following relationship:

Expected non-HIV MMR =EPMNA*(1-a)*E/B

Where:

a = the proportion of HIV-related deaths among all deaths to women aged 15–49 years

E = the total number of deaths to women of reproductive age

B = the number of births.

Estimation of HIV-related indirect maternal deaths:

For countries with generalized HIV epidemics and high HIV prevalence, HIV/AIDS is a leading cause of death during pregnancy and post-delivery. There is also some evidence from community studies that women with HIV infection have a higher risk of maternal death, although this may be offset by lower fertility. If HIV is prevalent, there will also be more incidental HIV deaths among pregnant and postpartum women. When estimating maternal mortality in these countries, it is, thus, important to differentiate between incidental HIV deaths (non-maternal deaths) and HIV-related indirect maternal deaths (maternal deaths caused by the aggravating effects of pregnancy on HIV) among HIV-positive pregnant and postpartum women who have died (i.e. among all HIV-related deaths occurring during pregnancy, childbirth and puerperium).

The number of HIV-related indirect maternal deaths D H I V , is estimated by:

D H I V = a ∙ E ∙ v ∙ u

Where:

a*E = the total number of HIV-related deaths among all deaths to women aged 15–49.

v = is the proportion of HIV-related deaths to women aged 15–49 that occur during pregnancy. The value of v can be computed as follows: v = c k GFR / [1 + c(k-1) GFR] where GFR is the general fertility rate, and where c is the average exposure time (in years) to the risk of pregnancy-related mortality per live birth (set equal to 1 for this analysis), and where k is the relative risk of dying from AIDS for a pregnant versus a non-pregnant woman (reflecting both the decreased fertility of HIV-positive women and the increased mortality risk of HIV-positive pregnant women). The value of k was set at 0.3.

u = is the fraction of pregnancy-related AIDS deaths assumed to be indirect maternal deaths. The United Nations Maternal Mortality Estimation Inter-Agency Group (UN MMEIG)/TAG reviewed available study data on AIDS deaths among pregnant women and recommended using u = 0.3.

For observed PMs, we assumed that the total reported maternal deaths are a combination of the proportion of reported non-HIV-related maternal deaths and the proportion of reported HIV-related (indirect) maternal deaths, where the latter is given by a*v for observations with a “pregnancy-related death” definition and a*v*u for observations with a “maternal death” definition.

4.d. Validation

Estimates are reviewed with Member States through a World Health Organization (WHO) country consultation process and SDG focal points. In 2001, the WHO Executive Board endorsed a resolution (EB. 107.R8) seeking to “establish a technical consultation process bringing together personnel and perspectives from Member States in different WHO regions”. A key objective of this consultation process is “to ensure that each Member State is consulted on the best data to be used”. Since the process is an integral step in the overall estimation strategy, it is described here in brief.

The country consultation process entails an exchange between WHO and technical focal person(s) in each country. It is carried out prior to the publication of estimates. During the consultation period, WHO invites focal person(s) to review input data sources, methods for estimation and the preliminary estimates. Focal person(s) are encouraged to submit additional data that may not have been taken into account in the preliminary estimates.

4.e. Adjustments

Full details on adjustments and formulas are published/available here:

(1) Peterson E, Chou D, Gemmill A, Moller AB, Say L, Alkema L. Estimating maternal mortality using vital registration data: a Bayesian hierarchical bivariate random walk model to estimate sensitivity and specificity of reporting for population-periods without validation data. 2019 (https://arxiv.org/abs/1909.08578)

(2) Trends in maternal mortality 2000 to 2020: estimates by WHO, UNICEF, UNFPA, World Bank Group and UNDESA/Population Division. Geneva: World Health Organization; 2023.2019 (https://www.who.int/reproductivehealth/publications/maternal-mortality-2000-2017/en/).

To summarise the key adjustments in brief:

  • Adjustments for variation in definitions of the input data:

Previous studies found incidental or accidental deaths (comprise 10% of pregnancy-related deaths (excluding HIV-related deaths) in sub-Saharan African countries, and 15% in other low- and middle-income countries. Adjustments are applied to pregnancy-related deaths to account for these non-maternal deaths.

  • Adjustment for crisis years:

The proportion of pregnancy-related deaths among the deaths attributable to mortality shock from crisis is assumed to be equal to the proportion of women in the population who are pregnant or postpartum at the time of the crisis. The proportion of pregnant women in the population is set equal to the general fertility rate, based on the assumption of a one-year period associated with a live birth. Additional uncertainty is added to the estimates of crisis years.

  • Adjustment for age distribution in population-based surveys:

Population-based surveys such as Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS) obtain information by interviewing respondents about the survival of their siblings. This approach, commonly referred to as the direct sisterhood method. Given the study design (based on sisters of respondents), the population exposed to risk may be atypical of the population at large. Therefore, we compute an age-standardized value of PM, based on the female population of households at the time of the survey.

  • Adjustment for underreporting (unregistered) and misclassification in civil registration and vital statistics (CRVS) systems:

Underreporting and misclassification in CRVS systems are accounted for with specialized studies. Model estimated country-year specific adjustment factors are obtained and applied to CRVS data.

  • Adjustment for under-reporting in non-CRVS and non-specialised sources:

It is widely believed that some form of upward adjustment is required for data sources that are not CRVS or specialised studies, to account for deaths early in pregnancy that might not have been captured. Therefore, an upwards adjustment of 10% was applied to maternal deaths that were not obtained from CRVS systems or specialized studies.

4.f. Treatment of missing values (i) at country level and (ii) at regional level

  • At country level

Missing values are treated at the country-level. This is done as follows. There is no treatment of missing values at the regional level.

Predictor variable data:

Complete and comparable predictor data is obtained by constructing time series estimates for predictor variables (covariates).

  • Gross domestic product (GDP) per capita, measured in purchasing power parity (PPP) equivalent international dollars using 2017 as the baseline
  • General fertility rate (GFR)
  • Skilled birth attendant (SBA)

Response variable data:

All-cause deaths for WRA, used to denominate maternal deaths in the statistic PM, are imputed when missing and in some cases overwritten.

  • Estimated all-cause deaths from by UNDP’s World Population Prospects 2022 lifetables were used to impute and overwrite all-cause deaths in specialized studies in which the search went beyond registration systems.
  • Civil Registration and Vital Statistics (CRVS) reported all-cause deaths were used to impute missing all-cause deaths in specialized studies in which the search was within registration systems.
  • Estimated all-cause deaths from by UNDP’s World Population Prospects 2022 Estimates were used to impute missing all-cause deaths in miscellaneous studies.

4.g. Regional aggregations

Regional aggregations are calculated by aggregating the national-level estimates. The size of a country is determined by the live births estimated by World Population Prospects. Aggregations are currently made for each of the UN Agencies that comprise the UN MMEIG.

4.h. Methods and guidance available to countries for the compilation of the data at the national level

The methodology used by countries to compile the data depends on the source input type (CRVS, specialised study etc). Useful references include:

Support and guidance to national authorities may also be requested from the WHO Secretariat.

4.i. Quality management

For information on data quality management, assurance, and assessment processes at WHO, please refer to: https://www.who.int/data/ddi

4.j. Quality assurance

For information on data quality management, assurance, and assessment processes at WHO, please refer to: https://www.who.int/data/ddi

4.k. Quality assessment

For information on data quality management, assurance, and assessment processes at WHO, please refer to: https://www.who.int/data/ddi

5. Data availability and disaggregation

Data Availability

Data availability is presented by country with the country profiles, please see here: https://www.who.int/data/gho/data/themes/maternal-and-reproductive-health/maternal-mortality-country-profiles

https://www.who.int/publications/i/item/9789240068759

Disaggregation:

Current maternal mortality ratio (MMR) estimates are reported at national, regional, and global levels. Countries and territories included in the analyses are WHO Member States with populations over 100 000, plus two territories (Puerto Rico, and the occupied Palestinian Territory, including east Jerusalem).

The time series available is currently 2000 to 2020.

6. Comparability/deviation from international standards

Sources of discrepancies:

The maternal mortality ratio is defined as the number of maternal deaths divided by live births. However, to account for potential incompleteness of death recording in various data sources, the United Nations Maternal Mortality Estimation Inter-Agency Group (UN MMEIG) first computes the fraction of deaths due to maternal causes from original data sources (referred to as the “proportion maternal”, or PM), and then applies that fraction to WHO estimates of total deaths among women of reproductive age to obtain an estimate of the number of maternal deaths.

In other words, the following fraction is first computed from country data sources:

PM= Number of maternal deaths 15-49/All female deaths at ages 15-49

and then the PM is used to compute the MMR as follows:

MMR=PM × (All female deaths at ages 15-49/Number of live births)

Where the estimate of all deaths at ages 15-49 in the second equation is derived from WHO Global Health Estimates life tables, and the number of live births is from the World Population Prospects 2019.

With this as background, a few reasons that MMEIG estimates may differ from national statistics are as follows:

  1. Civil registration and vital statistics systems are not always complete (i.e., they do not always capture 100% of all deaths) and completeness may change over time. The MMEIG estimation approach attempts to correct for this by using the above approach, which involves first computing the PM.
  2. The MMEIG often applies adjustment factors to the PM computed from original data to account for measurement issues (such as how the country defined “maternal” deaths; misclassification; or incompleteness).
  3. The MMEIG uses the standardized series of live births from the United Nations Population Division, as published in World Population Prospects 2022, in the denominator of the MMR equation. To better inform the WPP, countries should discuss discrepancies directly with the United Nations Population Division: the contact address is population@un.org; this email address is monitored regularly, and messages are dispatched to the appropriate analysts for each country or concern.
  4. Statistically speaking, maternal deaths are a relatively rare event, which can lead to noisy time trends in data over time. As the goal of the MMEIG estimates is to track long term progress in reducing maternal mortality, the estimation process involves some smoothing to generate a curve that better captures changes in underlying risk

7. References and Documentation

URL: https://www.who.int/publications/i/item/9789240068759

References:

  1. Trends in maternal mortality 2000 to 2020: estimates by WHO, UNICEF, UNFPA, World Bank Group and UNDESA/Population Division. Geneva: World Health Organization; 2023.
  2. Peterson E, Chou D, Gemmill A, Moller AB, Say L, Alkema L. Estimating maternal mortality using vital registration data: a Bayesian hierarchical bivariate random walk model to estimate sensitivity and specificity of reporting for population-periods without validation data. 2019 (https://arxiv.org/abs/1909.08578).