Risk and timing of miscarriage and stillbirth in five low- and middle-income countries: evidence from longitudinal cohort studies

This longitudinal study of over 5,700 pregnant women across five low- and middle-income countries reveals that pregnancy loss remains a significant public health challenge with a cumulative risk of 103 per 1,000 pregnancies after 8 weeks, peaking in early gestation and among women over 35, while highlighting that conventional estimates often underestimate rates due to delayed enrollment.

The Gist

Imagine pregnancy as a long, winding journey up a mountain. For most travelers, the path is clear, and they reach the summit (the birth of a healthy baby) safely. But for some, the journey ends before they get there. Sometimes the path disappears early on (a miscarriage), and sometimes the traveler makes it most of the way up but slips near the top (a stillbirth).

This paper is like a team of explorers who decided to track 5,755 women on this mountain journey across five different countries in Africa and India (Côte d'Ivoire, Ethiopia, Kenya, India, and South Africa). Their goal was to figure out exactly where on the mountain these accidents happen most often and why.

Here is the story of what they found, told in simple terms:

1. The "Late Arrival" Problem

In the past, researchers tried to count these mountain accidents by asking women at the top of the mountain (or near the end of their pregnancy) if they had lost a baby earlier.

• The Flaw: This is like asking a marathon runner at the finish line, "Did you trip in the first mile?" Many runners who tripped early didn't even make it to the finish line to answer the question.
• The Fix: This study was different. They started tracking women as soon as they showed up at the clinic, even if it was early in the pregnancy. They followed them all the way through.
• The Result: Because they looked at the whole journey, not just the end, they found that pregnancy loss is much more common than we thought. Previous estimates were like looking at a map with half the roads missing.

2. The "Danger Zones" on the Mountain

The researchers mapped out exactly when the risks are highest. They found two main danger zones:

• Zone 1: The Early Climb (Weeks 8 to 16): This is the steepest, most dangerous part of the climb. The risk of losing a pregnancy is highest here. It's like the rocky, unstable ground right after the trailhead.
• Zone 2: The Summit Approach (After Week 36): There is a second, smaller spike in danger right near the end of the journey. This suggests that even when a baby is almost ready, things can go wrong if the final stretch isn't managed carefully.

The Big Numbers:
If you follow 1,000 pregnancies that make it past the first 8 weeks:

• About 84 will be lost before week 28 (miscarriage).
• About 19 will be lost between week 28 and the end (stillbirth).
• Total: About 103 out of every 1,000 pregnancies end in loss. That's roughly 1 in 10.

3. Who is at Risk?

The team looked at the "travelers" to see if their background made the journey harder.

• Age: Women over 35 faced a significantly higher risk (almost double) of losing the pregnancy. It's like an older climber might find the steep rocks more difficult to navigate.
• Money and Education: Surprisingly, being poor or having less education didn't consistently predict who would lose a baby in this specific study. This suggests that the risks are often biological or related to the health system itself, rather than just personal circumstances.
• Location Matters: The journey was much riskier in Côte d'Ivoire (where the loss rate was highest) compared to South Africa (where it was lowest). This implies that the "trail conditions" (healthcare quality, local diseases, nutrition) vary wildly from country to country.

4. Why This Matters

Think of pregnancy loss as a silent storm. Because many of these losses happen very early (before a woman even knows she is pregnant or before she sees a doctor), they often go uncounted.

• The Old Way: We were only counting the storms that happened near the end of the trip.
• The New Way: This study shows us that the storm is actually raging right at the start of the trip, too.

The Takeaway

The authors are saying: "We need to change our map."
We can't just focus on helping women during the final weeks of pregnancy or during delivery. We need to start helping them much earlier.

• We need better care in the first few months.
• We need to understand why some countries have much rougher "trails" than others.
• We need to stop underestimating how common these losses are.

If we want to reach the global goal of ending preventable deaths by 2035, we have to acknowledge that the danger starts at the very bottom of the mountain, not just at the top.

Technical Summary

1. Problem Statement

Pregnancy loss (miscarriage and stillbirth) represents a significant public health burden with profound physical, psychological, and economic consequences for women and families. However, accurate prevalence estimates in Low- and Middle-Income Countries (LMICs) are scarce due to:

• Data Gaps: National health surveys often lack miscarriage data, and vital registration systems are incomplete.
• Methodological Bias: Existing studies frequently rely on retrospective surveys, facility-based data, or late pregnancy enrollment. This leads to systematic underestimation of miscarriage rates because pregnancies lost before enrollment or before the first antenatal care (ANC) visit are excluded.
• Lack of Longitudinal Data: Few multi-country studies utilize prospective longitudinal designs to capture gestation-specific patterns of loss.

2. Methodology

Study Design and Setting

• Data Source: The study analyzed longitudinal data from 5,755 pregnant women across five countries: Côte d'Ivoire, Ethiopia, India, Kenya, and South Africa.
• Cohorts: Data was drawn from the Maternal and Neonatal Health eCohort Study (MNH eCohort) and the Multigenerational Birth Cohort in Taabo (Taabo MGC).
• Recruitment: Women were enrolled during their first ANC visit (MNH eCohort) or upon pregnancy registration in a surveillance system (Taabo MGC) and followed through delivery.

Definitions and Outcomes

• Primary Outcomes:
  • Miscarriage: Fetal death before 28 weeks of gestation.
  • Stillbirth: Fetal death at 28 weeks or later.
• Secondary Definitions: Analyses were also conducted using clinical thresholds (early miscarriage <13 weeks, late miscarriage 13–24 weeks, stillbirth ≥24 weeks) to facilitate international comparison.
• Covariates: Maternal age, education level, household wealth index (asset-based), and location (urban/rural).

Statistical Analysis

• Competing Risks Survival Analysis: The authors employed a competing-risks framework where live birth is treated as a competing event rather than a censoring event. This prevents the upward bias in risk estimates that occurs when standard survival methods assume non-informative censoring.
• Adjustment for Delayed Entry: The models explicitly accounted for delayed enrollment (women entering the study in the 2nd or 3rd trimester), a critical factor often ignored in crude estimates.
• Risk Estimation:
  • Cumulative Incidence Functions (CIFs): Used to estimate cumulative risks of miscarriage (by 28 weeks) and stillbirth (28–44 weeks).
  • Fetuses-at-Risk Approach: Calculated prospective risks in 4-week intervals among fetuses surviving to the start of the interval.
  • Regression Modeling: Fine and Gray sub-distribution hazard models were used to assess associations between maternal characteristics and pregnancy loss, adjusting for country fixed effects.

3. Key Contributions

1. Correction of Underestimation Bias: The study demonstrates that conventional estimates failing to account for delayed enrollment and loss-to-follow-up significantly underestimate miscarriage rates. By using survival analysis with competing risks, the authors provide a more accurate picture of early pregnancy loss.
2. Gestation-Specific Risk Profiles: The study maps the risk of loss across the entire gestation, revealing that risk is not uniform but peaks in early gestation and rises again late in pregnancy.
3. Multi-Country Comparative Data: It provides the first direct comparison of pregnancy loss risks across five diverse LMICs with varying economic and health system contexts, highlighting significant inter-country heterogeneity.
4. Methodological Rigor: The application of the "fetuses-at-risk" perspective and competing-risks models sets a new standard for analyzing pregnancy loss in settings where early detection is challenging.

4. Key Results

Prevalence and Timing

• Cumulative Risk (8–44 weeks): Among pregnancies surviving to 8 weeks, the total risk of pregnancy loss was 103 per 1,000 (95% CI: 88–119).
  • Miscarriage (8–28 weeks): 84 per 1,000 (95% CI: 69–100).
  • Stillbirth (28–44 weeks): 19 per 1,000 (95% CI: 15–24).
• Timing of Loss:
  • The highest risk occurs between 8 and 16 weeks of gestation.
  • A secondary rise in risk is observed after 36 weeks, suggesting post-term complications or intrapartum issues.
• Country Variations:
  • Highest Risk: Côte d'Ivoire (Cumulative loss: 138/1,000; Stillbirth: 30/1,000).
  • Lowest Risk: South Africa (Cumulative loss: 85/1,000; Stillbirth: 12/1,000).
  • Stillbirth Rates (per 1,000 live births): Ranged from 13 (South Africa) to 34 (Côte d'Ivoire).

Risk Factors

• Maternal Age: Women aged ≥35 years had a significantly higher hazard of pregnancy loss compared to those aged 20–34 (Hazard Ratio [HR] 1.78; 95% CI: 1.27–2.48).
• Socioeconomic Status: No consistent association was found between pregnancy loss and maternal education or household wealth across the pooled sample.
• Country Effect: Women in Côte d'Ivoire had nearly double the risk of loss compared to those in Ethiopia (HR 1.94), even after adjusting for individual characteristics.

Comparison with Literature

• The study's estimated miscarriage risk (84/1,000) is significantly higher than the pooled estimate from previous literature (56/1,000) and crude estimates from the same cohort (35/1,000), confirming that late enrollment leads to severe underestimation.
• Stillbirth estimates aligned closely with existing literature, as these events occur later in pregnancy and are less affected by enrollment timing.

5. Significance and Implications

• Public Health Burden: Pregnancy loss in LMICs is a major, underestimated burden. The findings suggest that global goals to reduce newborn deaths and stillbirths (e.g., Every Newborn Action Plan) may be off-track if early gestation losses are not addressed.
• Intervention Timing: Interventions focused solely on late pregnancy and delivery are insufficient. The data supports the urgent need for enhanced surveillance and interventions during early gestation (first and second trimesters).
• Policy Recommendations:
  • Early Antenatal Care: Strengthening systems to detect pregnancies earlier and provide timely risk screening.
  • Data Systems: Moving beyond retrospective surveys to longitudinal cohort designs to capture the full spectrum of pregnancy loss.
  • Context-Specific Strategies: Given the high variability between countries (e.g., Côte d'Ivoire vs. South Africa), interventions must be tailored to local health system capacities and risk profiles.

Conclusion: The study concludes that pregnancy loss remains a critical, under-measured public health issue in LMICs. Accurate measurement requires longitudinal data and advanced statistical methods to account for delayed entry, revealing that the true burden of preventable fetal loss is substantially higher than previously reported.