Simulations Demonstrate the Importance of Pooled Data in Environmental Health Studies

Recent research highlights the significant potential of combining data from multiple studies to better understand the health effects of environmental chemicals. Conducted by Columbia University’s Mailman School of Public Health and published in the American Journal of Epidemiology, the study addresses longstanding challenges in environmental epidemiology, where inconsistent findings have often hindered consensus on chemical health risks.

The core issue stems from limited exposure ranges within individual studies, resulting in underpowered analyses that struggle to detect true dose-response relationships. To illustrate this, researchers used simulated datasets to evaluate how well single and pooled studies can identify the connection between chemical exposure and health outcomes. Their findings underscore the value of data pooling, even when different cohorts have varying confounding factors.

Focusing on polychlorinated biphenyls (PCBs), a class of persistent environmental pollutants, the team examined the relationship between maternal PCB-153 levels in blood and birthweight. This association has shown inconsistent results in past research. The study utilized five hypothetical populations with diverse exposure distributions, based on real data from cohorts in New York City, Israel, and California. Simulating these environments revealed that individual studies often fail to capture the true relationship when exposure variability is limited. Conversely, combined data more accurately reflected the expected dose-response curve.

Lead researcher Eva Siegel emphasized that chemicals such as endocrine-disrupting POPs can affect health even at low doses. She stressed the importance of broader data collection across different populations to fully understand these effects. Senior author Pam Factor-Litvak added that harmonizing data across studies—even with varied confounders—is crucial to overcoming the limitations of narrow exposure ranges.

This research advocates for increased collaboration and data sharing among environmental health studies. Pooling data helps scientists detect subtle effects, especially at low exposure levels, ultimately advancing public health understanding of chemical risks.

Additional contributors include Matt Lamb, Jeff Goldsmith, Andrew Rundle, Andreas Neophytou, Matitiahu Berkovitch, and Barbara Cohn.

For more details, see the full study: Eva Laura Siegel et al., 'Using simulations to explore the conditions under which true dose-response relationships are detectable for environmental exposures: polychlorinated biphenyls (PCBs) and birthweight,' American Journal of Epidemiology, 2025. [DOI: 10.1093/aje/kwaf020]