Fuglsang 1989.

Methods	Single centre Blinding of randomization ‐ uncertain Blinding of intervention ‐ yes Complete followup ‐ no; 9 of the 27 randomized infants were excluded by day 5 because they had causes (6 with septicemia and need for assisted ventilation, 2 with cerebral hemorrhage) and 1 withdrawn by parents. Blinding of outcome measure ‐ yes, continuous polygraph research recording while treatment blinded.
Participants	Data were reported for 18 preterm infants with idiopathic apnea requiring treatment. Mean (SD) comparisons between groups; weeks of gestational age at birth 31 (3) in the caffeine group and 30 (2) in the theophylline group; birth weight 1499 (467) grams and 1351 (489) respectively, age in days at randomization 8 (11) and 7 (13).
Interventions	Caffeine citrate 20 mg/kg loading dose and maintenance dose of 5 mg/kg once daily Aminophylline used as theophylline 7.5 mg/kg loading dose and maintenance dose of 3.75 mg/kg 12 hourly. Ethylenediamine salt of theophylline (=aminophylline) was used. Both drugs were given orally.
Outcomes	Apnea ‐ cessation of breathing for more than 20 secs. In the paper the results were in a graph. The author provided us with a table of daily mean and SD of apnea per 24 hours for each group. Rates on day 2 (middle of 1‐3) and day 6 (middle of 5‐7) were used and converted to rates of apnea/100mins as in other studies. Bradycardia ‐ heart rate below 100 beats/min ‐ data not reported in the paper Side effects ‐ raised mean heart rate, excessive central nervous system stimulation or dyspepsia. These were not reported as individual data but verbal comments in the text of the paper indicated there was no side effects in either group.
Notes	In the first publication of this review and in updates this trial remained excluded because of the large proportion of infants excluded (9). The author provided valuable information and the baseline comparison of the remaining two groups (9 infants in each) was similar.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Unclear
Blinding (performance bias and detection bias) All outcomes	Low risk	Blinding of interventions and outcome assessments ‐ stated in paper and from author
Incomplete outcome data (attrition bias) All outcomes	Low risk	9 of the 27 randomized infants were excluded by day 5 because they had causes (6 with septicemia and need for assisted ventilation, 2 with cerebral hemorrhage) and 1 withdrawn by parents
Selective reporting (reporting bias)	High risk	As stated above in incomplete outcome data addressed
Other bias	Unclear risk	Not stated