524. How Important Is Breastfeeding, Really?

Episode Summary

Executive Summary: This episode examines the challenge of proving whether breastfeeding causes better child outcomes. Using a weekend-vs-weekday natural experiment in the UK, economist Emla Fitzsimons finds breastfeeding support affects breastfeeding rates among disadvantaged mothers and that breastfeeding for at least 90 days improves children’s cognition, especially expressive language, through age five. Emily Oster argues the overall effects are modest and parents need less pressure, more support.

Main Topics: Why breastfeeding is hard to study causally (Priority: 5/5): The episode frames breastfeeding as a classic case where observational studies are confounded by family background, income, education, and support differences, making associations difficult to interpret as cause and effect. Emily Oster on data, parenting, and research quality (Priority: 4/5): Oster explains why parenting research often relies on flawed comparisons and why better methods are needed. She argues breastfeeding decisions should be guided by family fit rather than moral pressure. Breastfeeding rates and policy pressure over time (Priority: 4/5): The show reviews the rise in breastfeeding in the U.S. from about 22% in 1972 to over 80% of infants starting with some breast milk by 2019, alongside changing guidelines and social norms. Weekend birth as a natural experiment (Priority: 5/5): Fitzsimons uses the fact that low-risk births on weekends in the UK receive less lactation support to test whether breastfeeding support changes breastfeeding behavior and later child outcomes. Cognitive benefits versus health outcomes (Priority: 5/5): The study finds breastfeeding benefits are concentrated in cognition, especially expressive language, with no clear effects on broader health or non-cognitive outcomes in this sample. Inequality and targeted support (Priority: 4/5): Effects are observed mainly among lower-educated, disadvantaged mothers, suggesting breastfeeding support may reduce educational and cognitive inequality if accessible. Policy and messaging implications (Priority: 4/5): Both guests stress that pressure to breastfeed can harm maternal mental health, but support should be expanded for those who want to breastfeed.

Key Arguments: Most breastfeeding research is correlational; mothers who breastfeed differ systematically from those who do not, so outcomes may reflect background advantages rather than breastfeeding itself. Oster argues better studies, including randomized trials and sibling comparisons, show only small impacts overall, with some short-term health benefits but little evidence for major long-term cognitive effects. Fitzsimons shows that weekend births reduce breastfeeding among low-educated mothers because hospital lactation support is less available, creating a quasi-random variation in breastfeeding exposure. Her study suggests breastfeeding for at least 90 days improves cognitive development, especially expressive language, with effects lasting to at least age five. The benefits appear concentrated among disadvantaged mothers, implying that support services can influence inequality in child outcomes. Maternal attachment does not appear to differ by breastfeeding status in Fitzsimons’s data, suggesting the observed cognitive effect is not explained by stronger maternal bonding. Both guests argue that anti-formula or pro-breastfeeding pressure should be replaced with practical support for families who choose breastfeeding.

Data Points: U.S. breastfeeding rate in 1972: 22% - After decades of decline, only about 22% of women breastfed their babies in 1972. Infants starting with some breast milk in 2019: more than 80% - By 2019, over 80% of infants in the U.S. started out receiving some breast milk. AAP exclusive breastfeeding recommendation: at least 6 months - In the late 1990s, the American Academy of Pediatrics recommended exclusive breastfeeding for at least six months. Weekend breastfeeding gap among low-educated mothers: 6 percentage points - A baby born on a Saturday was six percentage points less likely to be breastfed than one born on a Monday in the disadvantaged sample. Weekend vs Monday breastfeeding example: 27% vs 21% - Illustrative comparison of breastfeeding rates for Saturday versus Monday births among the relevant group. Sample size in UK cohort: over 19,000 children - Fitzsimons used the Millennium Cohort Study tracking children born in the UK around the turn of the millennium. Breastfeeding exposure threshold studied: at least 3 months / 90 days - The causal analysis examined breastfeeding for at least three months versus less than that. Effect size on expressive language: 15 percentile points - Breastfeeding for at least 90 days shifted children up 15 percentile points in expressive language distribution. Age of observed cognitive effects: through at least age five - The strongest effects were observed to persist at least until age five. Age range of broader child outcomes studied: up to age seven - Health, socio-emotional, and behavioral outcomes were assessed through age seven. Education follow-up available: up to age 16 - The researchers noted they now have educational attainment data through age 16 for future analysis. Randomized trial referenced: Probit trial in Belarus, 1990s - Oster cited a large randomized encouragement trial as one of the strongest prior studies.

Pivotal Quotes: "What really knew and what was actually causal versus more sort of observational association type studies." — Emla Fitzsimons: Explaining why she studied breastfeeding after noticing reduced weekend hospital support for new mothers. "Some of the ways we have been pressuring people to breastfeed have been counterproductive and can actually be quite harmful." — Emily Oster: On the risks of breastfeeding messaging that creates guilt and stress for parents. "We found that all of the effects appears concentrated on children's cognition, and we observed strong effects lasting at least to age five." — Emla Fitzsimons: Summarizing the main finding of the weekend-support natural experiment.

Implications: Breastfeeding likely helps some children cognitively, especially when support enables disadvantaged mothers to persist. But the episode argues against guilt-based messaging: families need access to support, not pressure, and future policy should target help where it changes behavior most.