Introduction: Respiratory support use in neonatal intensive care units (NICUs) varies worldwide, influenced by clinical practices, resources, and patient populations. Whether high-altitude independently affects the duration of respiratory support in preterm infants remains unclear. Objective: To determine whether altitude is independently associated with the duration of respiratory support in preterm infants ≤32 weeks’ gestational age admitted to Latin American NICUs. Methods: We performed a multicenter observational cohort study by secondary analysis of prospectively collected data from the EpicLatino Network, a registry of NICUs across Latin America (2015–2022). Infants ≤32 weeks who received invasive or non-invasive respiratory support were included; supplemental oxygen delivered via low-flow nasal cannula or oxygen hood was not considered respiratory support, and those with missing outcome data were excluded. The primary outcome was total duration of respiratory support, measured as total days of support until discontinuation, discharge, transfer, or truncation by death. Altitude was classified as high (≥2000 m) or low (<2000 m). Multivariable analyses were adjusted for neonatal, maternal, and unit characteristics. Results: A total of 4,428 infants were included; 2,723 (61.5%) in low-altitude NICUs and 1,705 (38.5%) in high-altitude NICUs. Overall, 81.4% discontinued respiratory support and 18.6% died. Mortality was 19.1% in low-altitude and 17.9% in high-altitude NICUs. Median duration of support was 8 days (IQR 5–14) overall, with 9 days (IQR 4–27) in low-altitude and 7 days (IQR 3–17) in high-altitude NICUs. High-altitude centers showed shorter respiratory support in unadjusted analyses. After adjustment for neonatal, maternal, and unit factors, altitude was not independently associated with support duration. Conclusions: After adjustment for neonatal, maternal, and unit factors, altitude was not independently associated with the duration of respiratory support. Importantly, high altitude was never associated with worse outcomes.
Hoyos et al. (2026) studied this question.