TABLE 2.

Summary of retrospective studies, post hoc analyses, randomised controlled trials (RCTs) and systematic reviews and meta-analyses comparing high versus low/standard doses of caffeine citrate in preterm infants

First author, year [ref.]	Study characteristics Patient characteristics Limitations	Regimen		Main significant findings
		High caffeine dose	Standard/low caffeine dose	Benefits of high caffeine dose	Drawbacks or no effects of high caffeine dose
Romagnoli, 1992 [88]	Single-centre RCT 37 total neonates, 14 (controls) versus 13 versus 10 neonates, born <32 GW Single centre; small sample size; unclear risk of most biases with incomplete outcome data	Group I: LD 10 mg·kg−1; MD 5 mg·kg−1	Group II: LD 10 mg·kg−1; MD 2.5 mg·kg−1	Decrease in the number of apnoeic spells in both treated groups compared with a control group (p<0.01)	Significantly lower frequency of side-effects such as tachycardia (p<0.001) and gastrointestinal intolerance in the low-dose group (nonsignificant)
Scanlon, 1992 [80]	Single-centre RCT 44 total neonates, 14 versus 16 neonates (14 infants treated with theophylline), born <31 GW, with frequent apnoeic attacks (≥10 in 8 h or 4 in 1 h) Single centre; small sample size; unclear risk of most biases with incomplete outcome data	LD 50 mg·kg−1; MD 12 mg·kg−1	LD 25 mg·kg−1; MD 6 mg·kg−1	Number of apnoea events·day−1 reduced by 1/3 within 24 h by standard dose treatment versus a reduction by >50% by the higher dose treatment within the same time period
Steer, 2003 [81]	Single-centre RCT 45 versus 40 versus 42 neonates <32 GW ventilated for >48 h Single centre; small sample size	High dose: LD 60 mg·kg−1; MD 30 mg·kg−1 Moderate dose: LD 30 mg·kg−1; MD 15 mg·kg−1	LD 6 mg·kg−1; MD 3 mg·kg−1	Reduction in documented apnoea episodes (p<0.02); Trend to decrease in failure of extubation in the two highest dose groups (24% versus 25% versus 45%, p=0.06)
Steer, 2004 [82]	Multicentre RCT Total of 234 neonates, 113 versus 121 neonates, born <30 GW ventilated for >48 h; Data on long-term neurodevelopment to be considered with caution due to 18% loss at follow-up and not being the primary outcome	MD 20 mg·kg−1 before a planned extubation or 6 h within an unplanned extubation	MD 5 mg·kg−1 before a planned extubation or 6 h within an unplanned extubation	Reduced rate of extubation failure (15.0% versus 29.8%, RR 0.51; NNT 7) Reduction in documented apnoea episodes (4 (1–12) versus 7 (2–22), p<0.01) ignificant difference in duration of MV in infants <28 GW (mean 14.4 days versus 22.1 days, p=0.01)	No difference in mortality, major morbidities, severe disability
Gray, 2011 [89]	Multicentre RCT Total of 287 neonates, 120 versus 126 neonates, born <30 GW Some incomplete outcome data (e.g. age at starting treatment)	LD 80 mg·kg−1; MD 20 mg·kg−1	LD 20 mg·kg−1; MD 5 mg·kg−1	Significantly greater mean general quotient in the high-dose group (98.0±13.8 versus 93.6±16.5, p=0.048) Nonsignificant trend for benefit in the high-dose caffeine group for death or major disability (15.4% versus 24.2%; RR 0.75, 95% CI 0.49–1.14)	No difference in temperament and behaviour
Mohammed, 2015 [83]	Single-centre RCT 60 versus 60 neonates, born <32 GW Single centre; small sample size	LD 40 mg·kg−1; MD 20 mg·kg−1	LD 20 mg·kg−1; MD 10 mg·kg−1	Reduction in extubation failure (p<0.05) Reduction in frequency of apnoea (p<0.001)	Significant increase in episodes of tachycardia (p<0.05) No difference in the incidence of BPD No difference in the incidence of ROP, IVH, PVL or LOS
McPherson, 2015 [85]	Single-centre RCT Total of 74 neonates, 37 versus 37 neonates, born ≤30 GW Pilot study with small sample size only powered to detect differences in the primary outcome of microstructural brain development at term-equivalent age	LD 80 mg·kg−1 over a 36-h period (40–20–10); MD 10 mg·kg−1	LD 30 mg·kg−1 over a 36-h period (20–10); MD 10 mg·kg−1		Increased incidence of cerebellar haemorrhage in the high-dose group (36% versus 10%, p=0.03), more deviant neurological signs (p=0.04) at term-equivalent age No differences in diffusion measures at term-equivalent age and developmental outcomes at 2 years
Zhao, 2016 [90]	Single-centre RCT 164 total infants, 82 versus 82 neonates, born <32 GW Single-centre; possible selection, detection and reporting biases	LD 20 mg·kg−1; MD 15 mg·kg−1	LD 20 mg·kg−1; MD 5 mg·kg−1	Reduction in the frequency of apnoea (10 versus 18, p=0.009) Higher success rate of ventilator removal (85% versus 70%, p=0.015)	No significant difference in death during hospitalisation, CLD and duration of hospital stay No significant difference in tachycardia, irritability, difficulty in feeding, hyperglycaemia, hypertension, digestive disorders and electrolyte disturbances
Vliegenthart, 2018 [84]	Systematic review and meta-analysis including 6 RCTs with a total of 620 preterm infants; GA ≤32 GW Overall quality of the outcome measures (GRADE) considered low to very low due to imprecision and inconstancy of the effect estimates; small sample sizes of the included studies	LD 10–80 mg·kg−1; MD 5–30 mg·kg−1	LD 6–30 mg·kg−1; MD 2.5–20 mg·kg−1	In the subgroup analysis for therapy duration >14 days, significant reduction in the combined outcome of mortality or BPD at 36 weeks PMA (3 studies, 428 patients) (TRR 0.76, 95% CI 0.59–0.98) and in BPD rates alone (TRR 0.72, 95% CI 0.54–0.97) Reduction in extubation failure (TRR 0.51, 95% CI 0.37–0.70)	No difference in mortality at discharge or at 12 months Increased risk of tachycardia in the HD group (RR 3.39, 95% CI 1.50–7.64) No difference in NEC, SIP, ROP, IVH, hyperglycaemia. Considerations: no meta-analysis on differences in apnoea frequency due to diverse definition of the outcome No meta-analysis on duration of respiratory support due to data reported in IQR Inadequate power to detect small but clinical relevant differences Considerable differences in administered caffeine doses between studies
Brattström, 2019 [87]	Systematic review and meta-analysis including 6 RCTs with a total of 816 preterm infants (GA ≤32 GW); LD 20–80 mg·kg−1; MD 3–20 mg·kg−1 Low quality of evidence mainly due to imprecision of the estimates, few events, small sample sizes and the wide confidence intervals of the meta-analysis	LD >20 mg·kg−1; MD >10 mg·kg−1	Doses lower than the high-caffeine group	Reduction in BPD at 36 weeks PMA (RR 0.76, 95% CI 0.60–0.96) Fewer cases of extubation failure (as defined by study authors, RR 0.51, 95% CI 0.36–0.71) and apnoeas (mean difference −5.68, −6.15—5.22), and shorter duration of MV (mean difference −1.69, −2.13—1.25) in the HD group	No difference in mortality (RR 0.85, 95% CI 0.53–1.38) No difference in IVH ≥3 (RR 1.41, 95% CI 0.71–2.79)
Chen, 2018 [92]	Systematic review and meta-analysis including 13 RCTs with 1515 infants, GA <32 GW Variable maintenance doses within the high- and low-dose range; only few trials assessing outcomes such as extubation failure, frequency of apnoea, apnoea duration; most studies in Chinese with low quality	Variable LD MD 10–20 mg·kg−1	Variable LD MD 5–10 mg·kg−1	Higher efficacy rate in the HD group (RR 1.37, 95% CI 1.18–1.45) Higher success rate of ventilator removal (3 studies, RR 1.74, 95% CI 1.04–2.90) Lower extubation failure rate in the HD group (3 studies, RR 0.5, 95% CI 0.35–0.71) Lower frequency of apnoea and shorter apnoea duration in the HD group (MD −1.55, 95% CI −2.72–−0.39 and MD −4.85, 95% CI −8.29–−1.40) Lower incidence of BPD in the HD group (RR 0.79, 95% CI 0.68–0.91)	Higher incidence of tachycardia in the HD group (RR 2.02, 95% CI 1.30–3.12)

GW: gestational weeks; LD: loading dose; MD: maintenance dose; RR: risk ratio; NNT: number needed to treat; MV: mechanical ventilation; BPD: bronchopulmonary dysplasia; ROP: retinopathy of prematurity; IVH: intraventricular haemorrhage; PVL: periventricular leukomalacia; LOS: late-onset sepsis; CLD: chronic lung disease; GRADE: Grading of Recommendations Assessment, Development and Evaluation; PMA: post-menstrual age; TRR: typical risk ratio; HD: high dose; NEC: necrotising enterocolitis; SIP: spontaneous intestinal perforation; IQR: interquartile range; GA: gestational age.