Improving global maternal mortality outcomes
May 8, 2023—Zachary Ward, research scientist in the Center for Health Decision Science at Harvard T.H. Chan School of Public Health, was the lead author on two recent papers in Nature Medicine: one that predicted trends in global maternal mortality and a second that evaluated the effectiveness of interventions to curb maternal deaths.
Q: Tell us about this pair of studies.
A: The first study describes the development and calibration of a simulation model for global maternal health. Simulating individual women in 200 countries over their reproductive life cycle, the model accounts for historical trends starting in 1990 and makes projections to 2050. It finds that, although maternal mortality is decreasing globally, it’s not decreasing at a fast enough rate to meet the United Nations’ Sustainable Development Goals, whose 2030 target is to achieve a global maternal mortality ratio [MMR] of 70 maternal deaths per 100,000 live births, with no country above 140. The model finds that on current trends we’re looking at an average global MMR of 167 by 2030—quite a ways off the global target of 70— with almost 60 countries projected to be above 140, mainly in sub-Saharan Africa.
The second study uses this model to look at the potential impact of different maternal health policy interventions—what happens if we focus on one at a time, or what happens when we combine them? We find that we need a comprehensive strategy to achieve that goal of a global MMR of 70. Focusing on just facility-based interventions like improving quality of care is necessary but not sufficient. Having more women deliver in facilities is necessary but not sufficient. Those two together could make a large improvement, but we’re going to have to also include family planning interventions and community-based interventions like antenatal care and improving referral links from the community to high-quality facilities in cases of emergency.
Q: Why did you decide to create a new model for this research?
A: Previous models have looked at associations between maternal mortality and country-level factors like GDP. The obvious limitation is that GDP isn’t a policy lever. We’re not going to go to a Minister of Health and say, if you want to improve maternal mortality, you should really be increasing your country’s GDP.
Here, we’re using a fundamentally different modeling approach where we model the ‘natural history’ of disease. This kind of modelling is common in decision science to look at things like the impact of cancer screening, where you need to model 40 or 50 years of someone’s life and a how disease progresses over time to estimate the impact of different interventions over the life course. We’re using that approach here to simulate the reproductive life cycles of women.
The downside is that it’s very computationally intensive. This is the most complicated model I’ve ever built. Because there are so many aspects to maternal health, and because we simulate individual women in 200 countries, there are over 400,000 parameters in the model, which took almost a year to fit to empirical data. But the payoff is that once it is calibrated, we can use it for policy analysis. So, one pro is that we can use our model to look at realistic policy interventions. It opens up new possibilities for testing out strategies in silico before we try to implement them. The other pro is that we can account for countries’ different priorities. India, for example, has falling fertility rates. Family planning interventions would have a limited incremental impact there compared to somewhere like Afghanistan, where fertility rates are still quite high. Hopefully policymakers will find these country-specific results more helpful within their own contexts.
Q: Is there reason to feel optimistic about the trajectory of global maternal health?
A: If trends continue at the current rate, it doesn’t look like the U.N. target for maternal mortality for 2030 would be met even by 2050. So I think one main takeaway is that efforts need to be redoubled to accelerate improvements in maternal health—and that the current downward trend isn’t necessarily guaranteed to continue without sustained efforts. Things are going generally in the right direction, but more needs to be done.
It’s not like maternal mortality is a new disease. It’s not a type of cancer that’s hard to cure even in high-income settings. We have clinical interventions that are very effective at preventing maternal deaths. What we’re not as good at is the implementation side of things: getting a health system up and running, improving performance along the cascade of care to ensure that women are receiving quality care when they need it.
A next step for our research is to expand this model to account for costs. That’s a key component for policymakers to consider as well. The goal is to have health systems working, but how do we get there? It’s not like we can roll out all interventions at the same time. Once we bring in the cost side of things, we can help figure out a good order for interventions and prioritize cost-effective pathways for scale-up.
