Commentary: WHO and IHME Estimates – Global, Regional and National Causes of Maternal Death

Posted on

By: Affette McCaw-Binns, Professor of Reproductive Health and Epidemiology, University of the West Indies, Mona, Jamaica

In the same week IHME and the WHO provided the maternal health community with much on which to masticate; regional data on the causes of maternal death. IHME gives more detailed estimates, with Latin America and the Caribbean (LAC) divided into Andean LA, Caribbean, Central LA, Southern LA, and Tropical LA – but presented together by WHO. Whether this detail is more accurate is debatable with only nine estimated HIV-related maternal deaths for 2013 across the five sub-regions by IHME!

When one compares and contrasts both papers (WHO, for 2003-2009; IHME, representing 2013) the relative rankings and number of deaths for LAC emerge below. They disagree on the relative burden of the three leading causes, with better agreement on less common causes. Where they vary widely is on the numbers, albeit for different periods. WHO estimated almost twice as many haemmorhage deaths and 18 times as many HIV deaths as IHME. Nonetheless, these rankings are useful for countries which lack cause-specific data to inform interventions to work toward the goal of ending preventable maternal deaths.

Categories of maternal death in LAC IHME – 2013Ranking [estimated deaths] WHO – 2003-9Ranking [estimated mean/year]
Hypertensive disorders 1 [1407] 2 [1875]
Indirect causes, excluding HIV 2 [1073] 3 [1463]
Haemorrhage 3 [1058] 1 [2000]
Other direct causes 4 [1030] 4 [1125]
Abortion 5 [942] 5 [863]
Sepsis 6 [796] 6 [725]
Obstructed labour 7 [430] 7 [413]
HIV 8 [9] 8 [163]

Also, these data assume a linear decline in maternal mortality, which may not be accurate. The initial investment to set up health systems and change health seeking behaviour can be very high.  Once the system is functioning however rapid improvements can be realized, which will plateau until technology and training can address the more obstinate problems including the fertility transition.

The most interesting feature of both estimates is recognition of the growing burden of indirect deaths.  IHME also reported on late maternal deaths, 80% of which in Jamaica is due to the delayed effects of pre-existing medical conditions. The authors of the WHO estimates (Say et al)lamented the difficulty in coming to grips with the underlying causes of these deaths, a structural ICD-10 problem that hopefully ICD-11 will address.  Medical conditions, coded using ICD-MM/ICD-10 guidelines combine multiple causes whose clinical features and management vary widely – for example sickle cell disease is lumped with nutritional anaemias. While the latter can be managed in the community, the former frequently requires tertiary care.  Cardiac anomalies, ranging from congenital defects to coronary heart disease, likewise get rolled together.

Other interesting findings are:

  1. Relative increase in sepsis – 3rd leading global cause of maternal death – is this puerperal sepsis in HIV positive women? Is this a classification problem given ICD-MM or is it real? What strategies are needed to reverse this?  Is antibiotic resistance a contributing factor?  Are we failing to identify and appropriately treat HIV+ women? What about the growing C-section rate?
  2. Given that rates in some regions are <100/100,000, there is the need to explore strategies to bring these regions in line with the developed world.

These new estimates will probably not alter maternal health policy in Jamaica since our data shows our progress has stagnated for 15 years while IHME says our deaths are falling. Discrepancies aside, the estimates confirm for the international community that ending preventable maternal mortality means addressing indirect deaths as women delay childbearing into ages when non-communicable diseases (NCDs) emerge. Hopefully this will stimulate research into effective strategies to reduce these deaths.

Finally, these are best guesses imputed from better functioning health systems to less sophisticated locations with less access and quality care, whose reality might be quite different.  Going forward, Say et al have called for investment in developing locally owned information. This will enable countries and communities to tailor interventions to the disparities within countries, targeting their most vulnerable women who bear the highest burden of ill-health and poor outcomes. What a country does not measure, does not count. Thirty years after Nairobi, do women and their babies still not count?