Confounding biases in studies on early- versus late-caffeine in preterm infants: a systematic review

• Published in: Pediatric research Vol. 88; no. 3; pp. 357 - 364
• Main Authors: Nylander Vujovic, Sandra, Nava, Chiara, Johansson, Minna, Bruschettini, Matteo
• Format: Journal Article
• Language: English
• Published: New York Nature Publishing Group US 01-09-2020; Nature Publishing Group
• Subjects: Caffeine; Clinical Medicine; Klinisk medicin; Medical and Health Sciences; Medicin och hälsovetenskap; Medicine; Medicine & Public Health; Newborn babies; Pediatric Surgery; Pediatrics; Pediatrik; Premature babies; Systematic Review
• ISSN: 0031-3998; 1530-0447
• DOI: 10.1038/s41390-020-0757-1

Background
Caffeine is indicated for the management of apnoea of prematurity and extubation in preterm infants. Early initiation of caffeine administration has increased in the past decades with the purpose of reducing respiratory morbidity. However, there might be harms associated with this approach. This systematic review aims to assess whether early administration of caffeine reduces morbidity and mortality in preterm infants.
Methods
The methods were published in a preregistered protocol. The literature search was performed in February 2019 with no restrictions for language or publication date. Randomised controlled trials (RCTs) and cohort studies comparing early versus late caffeine administration to infants born before week 34 were included.
Results
Two RCTs and 14 cohort studies were included. All studies but one had a serious/critical overall risk of bias. Few studies reported on long-term or patient-relevant outcomes. No meta-analysis could be performed.
Conclusion
Based on the available evidence, no conclusions about the optimal timing of caffeine administration can be drawn. There are inherent methodological problems in the cohort studies. RCTs are needed to answer the question of optimal timing for caffeine administration in neonatal care. Future trials should focus on outcomes relevant to patients and their families and include long-term outcomes.