Physician Quality Shapes Birth Outcomes: Evidence from a Colombian Health Reform

A groundbreaking new study from the World Bank’s Development Impact Group, Harvard Business School’s Digital Reskilling Lab, Banco de la República de Colombia, and the University of California, Los Angeles reveals how the random assignment of newly graduated physicians dramatically affects birth outcomes in Colombia. The research leverages a natural experiment built into Colombia’s health system, where over 2,100 new doctors were randomly placed in 618 rural health centers as part of the country’s mandatory Servicio Social Obligatorio (SSO) program. This system provided an unprecedented opportunity to assess how the quality of doctors, measured through their scores on mandatory graduation exams, impacts the health of newborns. The study followed 255,089 births, and its findings are both robust and alarming: physician skill can be a life-defining variable for infants, even from their very first breath.

More Skill, Healthier Babies

The study defined an “unhealthy” baby as one born with any of three common complications: low birth weight (under 2,500 grams), prematurity (born before 37 weeks), or a low Apgar score (below 7). The researchers found that mothers treated by physicians who scored one standard deviation higher on graduation exams were 9.14 percent less likely to give birth to an unhealthy child. Specifically, such mothers were 9.57 percent less likely to have low-birth-weight babies, 10.99 percent less likely to have premature babies, and 11.56 percent less likely to deliver babies with low Apgar scores. These impacts are significant, especially in underserved areas where access to specialist care is extremely limited. Most LHCs in rural Colombia have only a handful of doctors, and the SSO physicians often serve as the sole providers of maternal care, including prenatal visits and childbirth services.

Not More Checkups, Smarter Ones

A common assumption in maternal health is that better outcomes come from more frequent checkups. However, the study reveals that the number of prenatal visits remained relatively unchanged regardless of the physician’s skill level. What distinguished higher-performing doctors was not quantity but quality, specifically, the targeting of care. Using machine learning tools like random forest and logistic regression, the researchers built models to predict which mothers were most at risk of delivering unhealthy babies, based on characteristics like age, marital status, education level, and previous childbirth history. They discovered that more-skilled physicians were significantly better at ensuring that high-risk mothers received the minimum recommended four prenatal checkups, in line with WHO standards. This better-targeted care occurred without reducing services for lower-risk mothers, indicating a sharper clinical prioritization by the more-skilled doctors.

A Rigorously Designed Natural Experiment

What makes this study particularly persuasive is its strong empirical strategy. The random assignment of physicians through the SSO program helped eliminate the usual biases caused by self-selection or administrative preference. To establish causality, the authors used an instrumental variable approach, exploiting the test scores of the first physician cohort a mother encountered during pregnancy to predict her overall exposure to physician skill. They also performed placebo tests using data from 2009–2012, before the implementation of the randomized system. These tests showed no significant correlation between physician quality and birth outcomes, strengthening the causal interpretation of the main results. Additionally, balance tests confirmed no systematic demographic or institutional differences across health centers based on physician skill levels.

Who Benefits the Most?

Although the improvements in birth outcomes were broadly shared, the study found that some groups seemed to benefit slightly more than others. The positive effects of skilled physicians were marginally greater among teenage mothers, first-time mothers, single mothers, and those with low education, demographics typically considered more vulnerable in maternal health literature. There was also a slightly stronger effect for male infants, who are biologically more susceptible to adverse conditions in utero. However, these subgroup differences were not statistically significant. The study also explored whether health centers with a higher proportion of SSO physicians saw stronger outcomes, and while the point estimates were indeed larger, the differences again lacked statistical significance. The key driver remained the skill level of the individual physician.

Policy Insights with Life-Altering Stakes

One of the most policy-relevant takeaways from the study is the predictive power of standardized test scores. The researchers found that these health-specific exam scores were strongly correlated with physician value-added, while other variables, such as gender, socioeconomic background, or even the ranking of the medical school attended, had little to no predictive value. This makes graduation exams a practical and scalable tool for identifying high-performing physicians. Policymakers could use this information to prioritize the deployment of more-skilled doctors to areas with higher maternal health risks, thereby improving health equity and outcomes without increasing healthcare costs.

The study offers some of the clearest evidence to date that physician competence, quantifiable and observable even before a doctor enters the workforce, can significantly influence neonatal health. It underscores the profound impact that effective human resource allocation can have in public health systems, especially in resource-constrained settings. By strategically placing more-skilled physicians where they are most needed, health systems can dramatically improve birth outcomes and give every child a better start in life, not by chance, but by design.