Humanosphere is on hiatus. Many thanks to our web design, development and hosting partner Culture Foundry for keeping the site active while we plan our next move. Culture Foundry builds, evolves and supports next-level websites and applications for clients you know, and you couldn’t ask for a better partner to help you thrive in digital. If you’re considering an ambitious website design or development project, we encourage you to make them your very first call.

WHO Responds on Maternal Deaths Data

I wrote last week about the World Health Organization revising down its numbers on maternal mortality, noting that the UN agency had arrived at basically the same place as some Seattle-based researchers roundly criticized for previously challenging WHO’s higher estimates.

Basically, it was good news. Maternal deaths are down to about 350,000 annually as opposed to the previous estimate of half a million. That’s still unacceptably high, of course, but it is progress.

Last spring, the New York Times quoted the editor of the Lancet, Richard Horton, saying he had been pressured “to delay or hold publication” of the Seattle group’s findings because some feared the lower numbers would hurt fund-raising and cause “potential political damage to maternal advocacy campaigns.”

WHO’s Colin Mathers was gracious enough to respond, explaining that they arrived at the same place as Seattle’s Institute for Health Metrics and Evaluation but by a different path, and with a few variations.

“Maternal mortality is one of the most difficult indicators to measure,” says Mathers. Causes of maternal death are often incorrectly reported, he says, and health institutions are often loathe to report them because of the “stigma” of having high rates of maternal deaths.

Mathers’ complete response is detailed and long. For you data geeks, who want to read the whole thing, please

Dear Tom,

The new figures are not the same as those from the Institute of Health Metrics. While the data used by IHME and the Interagency Group are similar, the methods and statistical models are somewhat different, leading to differences in individual country figures for some countries. In terms of results, estimated number of global maternal deaths for 2008 by the Interagency analysis (358 000) are very close to those reported by IHME (342 900), but the IHME estimated global total for 1990 was lower: 441 500 compared to 546 000 from the Interagency group. As a result, IHME estimated global trend of 1.4% average annual decline in maternal mortality ratio is lower than the 2.3% estimated by the Interagency analysis. IHME estimated a somewhat higher number of pregnancy-related HIV deaths in 2008 (61 400) than the Interagency Group (42 000 of which one half were considered to be incidental to pregnancy).

Maternal mortality is one of the most difficult indicators to measure reliably. Not only does it require a correct assessment of the number of deaths of women of reproductive age, which in itself may be a challenge in many developing countries, but it also requires the correct classification of these deaths by cause. Maternal deaths are more often misclassified than others, not only because they are easily confused with deaths due to other causes, but also because health institutions may prefer to attribute them to such other causes, due to the stigma of inadequate treatment associated with maternal death. Consequently, even in the best civil registration systems in the world, it has been found that maternal causes can be substantially under-reported. Given that different data sources will not necessarily yield the same results and that common adjustments to those data involve assumptions, variations in the estimates that can be produced for the same country can be considerable. Obviously this creates problems for both countries and for agencies as the estimates serve for monitoring and evaluation purposes that can have important political implications.

The country-level estimates of maternal mortality essentially fall into four categories, according to available sources of data and adjustments required for misclassification and under-reporting:

1. Countries with good quality death registration systems that capture essentially all deaths in the population. The issue for these countries is the level of misclassification of maternal deaths, particularly those indirect maternal deaths where pregnancy has exacerbated an existing illness. In the United Kingdom for example, the Confidential Enquiry into Maternal Deaths has regularly assessed that the total maternal deaths are around 1.4 to 1.9 times higher than the number recorded in death registration data (see the Interagency Report for details). The application of an adjustment factor based on these studies resulted in an estimated maternal mortality ratio of 12 for the UK in 2008. In comparison, IHME estimated an MMR of 8.2 for 2008 using different assumptions regarding the level of misclassification.

2. Countries with survey or census data on pregnancy-related deaths as reported by respondents. Such data are typically available only for a few points in time at most, and differences can arise in projected estimates for 2008 due to differences in statistical models and other assumptions. In particular, different assumptions and adjustments about the level of under-reporting (for example, women may not know that deaths of their siblings were related to pregnancy or abortion), and whether HIV or other disease deaths during pregnancy were incidental or indirect maternal deaths, may result in differences in estimates. Additionally, the methods used by both the Interagency Group and IHME involve estimating the proportion of all deaths in the reproductive age range that were due to maternal causes and then applying that estimate to separate estimates of total female death rates. The Interagency Group used all-cause adult mortality estimates published by the World Health Organization whereas IHME developed its own life tables.

3. Countries without any available data on levels of maternal mortality. For these countries, statistical models were used by both the Interagency Group and IHME to estimate likely levels of maternal mortality based on factors such as the level of development, the fertility rate and some measure of health system quality. These statistical models differed and resulted in differences in estimates for countries without data.For example, IHME estimated an MMR of 674 for Somalia in 2008 compared to 1200 by the Interagency Group.

Whatever differences between country estimates there are, the main point remains that maternal mortality rates are horrendously high in many low income countries, and that there are huge inequalities across regions of the world. The UN agencies are working together with academic experts to improve the measurement and estimation of maternal mortality and its trends, in order to monitor and highlight progress and lack of progress. The agencies also work closely with countries. All numbers go through a process of country consultation, in which new data are uncovered, estimates are discussed and improved as necessary. The aim is to come up with the best estimates that can be used by countries. We are welcoming the efforts of other groups to improve the estimation methods and are looking forward to collaborate with all interested parties to share data, sort out different methods and develop tools that can be used by countries.


Colin Mathers


About Author

Tom Paulson

Tom Paulson is founder and lead journalist at Humanosphere. Prior to operating this online news site, he reported on science,  medicine, health policy, aid and development for the Seattle Post-Intelligencer. Contact him at tom[at] or follow him on Twitter @tompaulson.