Neonatal TPN: the 3-day diabetes signal is not a stop rule

In a South Korean claims cohort of 2,303,567 first-year survivors, 7,656 had recorded neonatal TPN. The primary weighted analysis found no association with later metabolic disease, while an exploratory analysis found T2DM HR 2.14 (95% CI 1.34-3.41) after at least 3 recorded days. Residual illness imbalance and missing dose data make that duration a risk marker, not a causal threshold.

THE HOT TAKE

With 2.3 million infants, this study looks definitive until the exposure groups come into view. Only 7,656 infants had recorded neonatal TPN, and the primary weighted analysis found no association with later metabolic disease. The eye-catching HR 2.14 for type 2 diabetes after at least 3 recorded days came from a secondary model that cannot separate TPN duration from the illness that prolonged it.

WHAT THE COHORT ACTUALLY MEASURED

Investigators used South Korea's national health-insurance claims for live births from 2005 through 2010. They excluded infants with chromosomal abnormalities or inborn metabolic disease, 2,485 infants who died in the first year, and 31 with missing baseline data. The remaining first-year survivors were followed through 2023. Exposure meant at least one TPN prescription or procedure claim in the first 28 days; outcomes were later ICD-10 claims, not laboratory-adjudicated diagnoses.