Indications and timings for caffeine initiation in preterm infants

Caffeine is commonly used in preterm infants to improve respiratory function. However, the optimal strategies for caffeine initiation have not been defined.

To evaluate the relative benefits and harms of different indications and timings for caffeine initiation in preterm infants.

We searched MEDLINE, Embase, CENTRAL, and trial registries up to 23 April 2025. We searched conference abstracts, checked references of related systematic reviews and included studies, and searched for errata/retractions of included studies.

Randomized controlled trials (RCTs) including cluster-RCTs and quasi-RCTs comparing different indications for caffeine initiation in preterm infants: • within two hours of life (C2H) versus 2 to 24 hours; • within 72 hours of life (C72H) versus after 72 hours; • C72H versus treatment of symptomatic infants; • while on mechanical ventilation versus caffeine initiated at the time of extubation; • initiated with minimal symptoms versus with moderate-to-severe apnea.

Critical outcomes: all-cause mortality, chronic lung disease (CLD), and adverse events. Important outcomes: duration of mechanical ventilation (DMV) (days), duration of hospital stay (DHS) until discharge (days), apnea, intermittent hypoxemia (IH), acute kidney injury, severe intraventricular hemorrhage, cerebral palsy, developmental delay, intellectual impairment, blindness, and sensorial deafness.

We used the Cochrane Risk of bias tool 2.0.

We conducted meta-analyses using fixed-effect models to calculate risk ratios (RRs) with 95% confidence intervals (CIs) and mean difference (MD) for dichotomous and continuous outcomes, respectively. We summarized the certainty of evidence according to GRADE methods.

We included 11 RCTs enrolling 774 infants with gestational age ranging from 23 to 34 weeks. Studies were published between 2014 and 2023. The 11 studies compared caffeine initiated: • C2H versus 2 to 24 hours (3 studies); • C72H versus after 72 hours (3 studies); • C72H versus treatment of symptomatic infants (2 studies); • while on mechanical ventilation versus caffeine initiated at the time of extubation (3 studies). No studies compared caffeine initiated with minimal symptoms versus caffeine initiated with moderate-to-severe apnea. We identified seven ongoing studies.

C2H compared to caffeine initiated 2 to 24 hours: the evidence is very uncertain about the effect on CLD (RR 0.64, 95% CI 0.35 to 1.17; I² = 0%; 2 studies, 81 participants; very low-certainty evidence) and on DMV (MD 3.00, 95% CI -3.16 to 9.16; I² not applicable; 1 study, 21 participants; very low-certainty evidence). No studies reported two critical outcomes: all-cause mortality until hospital discharge and any adverse event. No studies reported DHS or apnea. C72H compared to caffeine initiated after 72 hours of life: the evidence is very uncertain about the effect on mortality (RR 2.00, 95% CI 0.19 to 21.36; I² not applicable; 1 study, 100 participants; very low-certainty evidence), DMV (MD 0.05, 95% CI -0.53 to 0.62; I² = 97%; 2 studies, 177 participants; very low-certainty evidence), and DHS (MD -4.59, 95% CI -10.51 to 1.33; I² not applicable; 1 study, 90 participants; very low-certainty evidence). C72H likely reduces CLD (RR 0.64, 95% CI 0.40 to 1.03; I² = 5%; 2 studies, 177 participants; moderate-certainty evidence) and apnea (number of infants with at least one episode, defined as interruption of breathing for more than 20 seconds until hospital discharge) (RR 0.58, 95% CI 0.37 to 0.90; I² = 81%; 2 studies, 177 participants; moderate-certainty evidence). No studies reported any adverse events. C72H for asymptomatic infants compared to treatment of symptomatic infants may result in little to no difference in all-cause mortality until hospital discharge (RR 0.95, 95% CI 0.47 to 1.92; I² not applicable; 1 study, 163 participants; low-certainty evidence) or in any adverse event leading to caffeine cessation until hospital discharge compared to caffeine treatment of symptomatic infants (RR 0.27, 95% CI 0.01 to 6.56; I² not applicable; 1 study, 163 participants; low-certainty evidence). It likely reduces CLD (RR 0.49, 95% CI 0.26 to 0.94; I² = 0%; 2 studies, 219 participants; moderate-certainty evidence). The evidence is very uncertain about DMV (MD -0.81, 95% CI -1.89 to 0.27; 1 study, 56 participants; very low-certainty evidence) and DHS (MD -1.43, 95% CI -4.90 to 2.04; 1 study, 56 participants; very low-certainty evidence). No studies reported apnea. Caffeine initiated during mechanical ventilation compared to caffeine initiated at the time of extubation may reduce CLD (RR 0.77, 95% CI 0.50 to 1.18; I² = 0%; 2 studies, 142 participants; low-certainty evidence) and may result in a large reduction in DMV (MD -1.90, 95% CI -3.14 to -0.66; I² not applicable; 1 study, 59 participants; low-certainty evidence). The evidence is very uncertain about the effect on all-cause mortality (RR 1.69, 95% CI 0.66 to 4.33; I² = 0%; 2 studies, 142 participants; very low-certainty evidence). No studies reported any adverse events, apnea, or DHS. No studies in any comparison reported the outcome IH.

The evidence is very uncertain about the effect of C2H and C72 on CLD and DMV compared to caffeine initiated within 24 hours of life and after 72 hours of life in high-risk preterm infants, respectively. The evidence is very uncertain about the effect of C72 on DHS. C72H likely reduces CLD and apnea. In high-risk preterm infants, C72H compared to caffeine treatment of symptomatic infants only, may result in little to no difference in all-cause mortality until hospital discharge, but likely reduces CLD. C72H of life may result in little to no difference in any adverse event leading to caffeine cessation. The evidence is very uncertain about the effect of C72H on DMV and DHS. The evidence is very uncertain about the effect on all-cause mortality until hospital discharge of caffeine initiated while on mechanical ventilation compared to initiated at time of extubation. Caffeine initiated while on mechanical ventilation may reduce CLD and may result in a large reduction in DMV. No studies reported any adverse events, apnea, or DHS. The results of ongoing studies might change the findings of this review.

This Cochrane review had dedicated funding: Forska Utan Djurförsök and ALF grant.

PROSPERO: crd.york.ac.uk/PROSPERO/view/CRD42024595715.