Is Early Caffeine Therapy Safe and Effective for Ventilated Preterm Infants?
-A   +A

​Is Early Caffeine Therapy Safe and Effective for Ventilated Preterm Infants?

Manuscript Citation: Amaro CM, Bello JA, Jain D, Ramnath A, D'ugard C, Vanbuskirk S, Bancalari E, Claure N. Early Caffeine and Weaning from Mechanical Ventilation in Preterm Infants: A Randomized, Placebo-Controlled Trial. The Journal of pediatrics. 2018 May 1;196:52-7.

Reviewed by:

Viral Jain1, Vivek Saroha2, Ravi Mangal Patel2, Alan Jobe1

1 - Perinatal Institute, Division of Neonatology, Cincinnati Children's Hospital, Cincinnati, OH, USA.

2- Division of Neonatal-Perinatal Medicine, Department of Pediatrics, Emory University School of Medicine, Atlanta, GA, USA

Type of Investigation: Treatment


Among preterm infants (23-30 weeks of gestation) requiring mechanical ventilation in the first 5 days of life, does receiving a 20 mg/kg loading dose followed by 5 mg/kg/day of caffeine, compared to treatment immediately before extubation, reduce the age at first successful extubation for > 24 hours, until 36 weeks postmenstrual age, discharge, or death, (which ever came first)?


Design: Allocation: Concealed; Blinding: Double - Blinded

Setting: Single, academically-affiliated NICU

Patients: (1) Inclusion Criteria: Infants between 23 - 30 weeks gestational age who required mechanical ventilation in the first 5 days of life. (2) Exclusion Criteria: Any infant with major congenital anomaly and small-for-gestation infant were excluded.

Intervention: Infants in the early caffeine group received a loading dose of caffeine citrate 20 mg/kg followed by a maintenance dose of 5 mg/kg/day, while infants in the control group received normal saline bolus and maintenance. Infants in the control group received a blinded loading dose of caffeine 20mg/kg before extubation, and vice versa infants in the early caffeine group received a blinded loading dose of saline (placebo). Randomization was stratified by gestational age group (23-26 or 27-30 weeks).


  • Primary outcome: Postnatal age of first extubation after which the infant remained extubated for >24 hours. If the infant remained extubated for > 24 hours, the study intervention was ended, and open label caffeine was given as per standard of care.

  • Secondary outcomes: (1) Total duration of mechanical ventilation (2) duration of supplemental oxygenation (3) incidence of BPD

Analysis and Sample Size: 110 infants were needed to detect a reduction in primary endpoint by >7 days, from a historical baseline of 15 ± 21 days, with 80% power and 5% significance. Kaplan-Meier log-rank test was used to compare age at first successful extubation.

Patient follow-up: Infants were followed until 36 weeks postmenstrual age, discharge, or death, which ever came first. Trial was stopped early at 75% enrollment (83 infants) by the data safety and monitoring board (DSMB) at second interim analysis for a persistent trend of higher mortality (although not statistically significant).

Main Results:

Baseline characteristicsEarly Caffeine (n=41)Control (n=42)
Male (%)30 (73)17 (41)
Gestational age (weeks)25.726.1
Birth weight (gms)670720
Age at randomization (hours ± SD)48 ± 2549 ± 31
5 minute Apgar Score <5 (%)10 (24)4 (9.5)

Study outcomesEarly Caffeine (n=41)Control (n=42)P
Median age in days at successful extubation (IQR)24 (10-41)20 (9-43)0.70
Median total duration of mechanical ventilation in days (IQR)32 (11-43)26 (10-44)0.72
Severe BPD (%)10 (30)12 (32)0.90
Death before discharge (n)9 (22)5 (12)0.22


Study Conclusion:

Early initiation of caffeine did not reduce the age of first successful extubation in preterm infants receiving mechanical intubation. However, the trial was stopped early as there was a non-significant trend toward higher mortality before discharge in the early caffeine group.


In the Caffeine for Apnea of Prematurity (CAP) trial, caffeine treatment, compared to placebo, reduced the risk of BPD and long-term motor impairment (1). A post-hoc analysis of the CAP trial found a greater benefit in reducing respiratory morbidity, including duration of ventilation, among infants who received early initiation of caffeine (<3 days of life), compared to later initiation (2). Thereafter, multiple observational studies have reported an association between early initiation of caffeine and a lower risk of BPD and shortened duration of ventilation (3). However, all these studies on early caffeine may have been biased by confounding by indication, with healthier infants receiving early caffeine therapy. 

Thus, the present randomized trial was designed to evaluate the safety and efficacy of early caffeine in decreasing the duration of mechanical ventilation for extremely preterm infants (23 to 30 weeks' gestational age). This study showed that the age at first successful extubation did not differ significantly between the early caffeine and control groups. Of note, the trial was stopped early because an interim analysis at 75% enrollment showed a potential difference in mortality before discharge, which upon final analysis was not significantly different among treatment arms (early caffeine vs. placebo: 22% vs 12%, relative risk 1.84; 95% CI 0.68-5.04; p=0.22). One prior observational study found an association between early initiation of caffeine and higher risk of mortality, although this association was attributed to survival bias (need to survive to receive later caffeine) (4). The present study also found no significant differences in secondary outcomes, including death before discharge, although was likely underpowered to detect clinically important differences in these outcomes.

Trials terminated early can be difficult to interpret. Stopping trials early, which is ethically appropriate to protect the safety of participants, can reduce power to detect clinically important differences in outcomes (5). In addition, trials stopped early may overestimate treatment effects (e.g. an observed effect on mortality), which could have diminished in magnitude had enrollment continued (5). This is because effects observed will fluctuate between larger random highs and lows early in the trial due to the small number of observations.

In addition, the interpretation of the outcomes in this trial is complicated by the presence of clinically important differences in baseline characteristics despite randomization. Due to the relatively small size of the trial and the outcome of chance, infants randomized to the early caffeine group had more male infants, a lower 5-minute Apgar score and were about 50 grams lower in birth weight, all important prognostic variables for survival that could have affected the results. However, in post-hoc analysis with adjustment for gestational age, 5-minute Apgar score and gender, there was no evidence of an effect of early caffeine on age at first successful extubation and on death, although estimates were not reported.

In conclusion, the results from this randomized trial do not support the prophylactic use of caffeine in mechanically ventilated infants at 23-30 weeks gestation to shorten the age at first successful extubation. These important findings contrast the results from observational studies of early caffeine therapy. Given the early termination of the trial, differences in prognostic variables for mortality between groups and the imprecision in the estimates of treatment effect of early caffeine on mortality (e.g. bounds of 95% confidence interval ranges from one-third reduced risk of mortality to 5-fold increased risk with early caffeine therapy), no confident conclusions can be determined from this study regarding the effect of early caffeine on mortality. To adequately assess the safety of early caffeine with regards to mortality, a very large multicenter trial would be needed to detect minimal clinically important differences in mortality. To our knowledge, no such trials are currently underway or planned. Thus, current evidence does not support the prophylactic use of early caffeine in preterm infants who are fully mechanically ventilated to facilitate early extubation.

EBM Lesson: Baseline imbalance in patient characteristics between treatment arms

Typically, Table 1 in a report of a randomized trial describes baseline characteristics of the population under investigation (often by treatment arm). These data can inform the reader about the external validity of the study population to their specific patient population of interest. It is likely that a small trial despite randomization may have differences in some baseline covariates between the treatment arms – an outcome of chance sometimes referred to as chance bias. As the sample size of a study increases, the absolute magnitude of any chance bias in outcome will tend to decrease (6). Evidence-based recommendations for reporting randomized trials such as the CONSORT statement discourage statistical testing and reporting of P values in comparisons of baseline covariates (7). Such significance tests assess the probability that observed baseline differences could have occurred by chance; however, we already know that any differences in baseline covariates after randomization (assuming it was performed properly) are due to chance (8). Stratifying randomization by important prognostic variables (as was done for gestational age in this study) can help reduce the probability of significant imbalances in key baseline characteristics. However, in the event of an imbalance of baseline covariates, whilst statistical testing for differences is discouraged, it is important to consider the magnitude of imbalance and resulting chance bias on the outcome. Post-hoc analyses can be used to adjust for imbalanced covariates to provide additional estimates of the treatment effect of an intervention that can be considered sensitivity analyses that support (or weaken) the findings of the primary analysis (9). Adjustment of prognostic covariates can also increase statistical power (10). The authors in this study addressed the imbalance in baseline covariates by performing post-hoc analyses adjusting for gender and 5 minute Apgar score. These analyses showed no effect of the treatment arm on either death or age at first successful extubation. This lesson illustrates how baseline differences in covariates should be reported (P values not necessary) and, when imbalances occur, how to consider these and account for their influence on estimates of treatment effects.


The Journal Club is a collaboration between the American Academy of Pediatrics - Section of Neonatal Perinatal Medicine and the International Society for Evidence-Based Neonatology (

Conflict of Interest:

 None declared


  1. Schmidt B, Roberts RS, Anderson PJ, Asztalos EV, Costantini L, Davis PG, et al. Academic Performance, Motor Function, and Behavior 11 Years After Neonatal Caffeine Citrate Therapy for Apnea of Prematurity: An 11-Year Follow-up of the CAP Randomized Clinical Trial. JAMA Pediatr. 2017;171(6):564-72.

  2. Davis PG, Schmidt B, Roberts RS, Doyle LW, Asztalos E, Haslam R, et al. Caffeine for Apnea of Prematurity trial: benefits may vary in subgroups. The Journal of pediatrics. 2010;156(3):382-7.

  3. Pakvasa MA, Saroha V, Patel RM. Optimizing Caffeine Use and Risk of Bronchopulmonary Dysplasia in Preterm Infants: A Systematic Review, Meta-analysis, and Application of Grading of Recommendations Assessment, Development, and Evaluation Methodology. Clin Perinatol. 2018;45(2):273-91.

  4. Dobson NR, Patel RM, Smith PB, Kuehn DR, Clark J, Vyas-Read S, et al. Trends in caffeine use and association between clinical outcomes and timing of therapy in very low birth weight infants. The Journal of pediatrics. 2014;164(5):992-8 e3.

  5. Viele K, McGlothlin A, Broglio K. Interpretation of Clinical Trials That Stopped Early. JAMA : the journal of the American Medical Association. 2016;315(15):1646-7.

  6. Roberts C, Torgerson DJ. Understanding controlled trials: baseline imbalance in randomised controlled trials. BMJ. 1999;319(7203):185.

  7. Schulz KF, Altman DG, Moher D. CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340.

  8. de Boer MR, Waterlander WE, Kuijper LD, Steenhuis IH, Twisk JW. Testing for baseline differences in randomized controlled trials: an unhealthy research behavior that is hard to eradicate. The international journal of behavioral nutrition and physical activity. 2015;12:4.

  9. Guideline on adjustment for baseline covariates in clinical trials. Committee for Medicinal Products for Human Use  In: CHMP, editor. London: European Medicines Agency (EMA). 2015.

  10. Kahan BC, Jairath V, Doré CJ, Morris TP. The risks and rewards of covariate adjustment in randomized trials: an assessment of 12 outcomes from 8 studies. Trials. 2014;15:139-.

            print           email           share