Sarrell 2006

Methods	Study design: randomized, double‐blind, parallel‐group trial. Study dates and duration: September 2003 to March 2004. Method of temperature measurement: rectal glass and mercury thermometer. Time points measured in study: daily temperature diary (parents asked to measure rectal temp at least 3 times daily during tx then once daily x 10 days), telephone interview at 24 hours and 48 hours, office visit day 3, 5, 10, and 14 day follow up evaluation.
Participants	Number: 480 Number of patients in each intervention: 160 in each of the 3 groups. Inclusion criteria: all consecutive children aged 6 ‐ 36 months who had rectal temperature ≥38.4 °C. Exclusion criteria: not attending daycare centers, had taken temperature‐altering drugs or antibiotics within 10 days before presentation, known abnormal liver or renal laboratory values, medical history of renal or hepatic impairment, gastrointestinal bleeding, known allergy to any antipyretic, congenital or acquired immunodeficiency, Reye syndrome, asthma, bronchiolitis or malignancy, and children whose caregivers were unable to apply the NCCPC to measure stress. Baseline characteristics: Age months (SD): paracetamol 18.6 (8.72), ibuprofen 19.5 (9.09), paracetamol plus ibuprofen 19.3 (9.29) Sex male (%): paracetamol 71 (46%), ibuprofen 73 (40%), paracetamol plus ibuprofen 62 (38%) Diagnoses: Paracetamol n(%)              Ibuprofen n(%)                Paracetamol plus Ibuprofen n(%) URI                    66(43)                81(52)                    80(51) AOM                 16(10)                 13(8)                       17(11) Pharyngitis         10(7)                  7(5)                           3(2) Bronchiolitis       8(5)                    7(5)                           9(6) Gastroenteritis    7(5)                    7(5)                           6(4) Viral illness        47(30)               40(25)                      40(26)
Interventions	Group1: paracetamol (12.5 mg/kg) q6h, max 50 mg/kg/day); half of the group received initial loading with paracetamol (25 mg/kg) and the other half received initial loading with ibuprofen (10 mg/kg) Group 2: ibuprofen (5 mg/kg) q8h, max 20 mg/kg/day; half of the group received initial loading with paracetamol (25 mg/kg) and the other half received initial loading with ibuprofen (10 mg/kg) Group 3: paracetamol (12.5 mg/kg/dose, max 50 mg/kg/d) alternating with ibuprofen (5 mg/kg/dose, max 20 mg/kg/d) q4h; half of the group received initial loading with paracetamol (25 mg/kg) and the other half received loading with ibuprofen (10 mg/kg)
Outcomes	Primary: Body temperature Stress score Amount of antipyretic used at the 3 day time point (number of doses) Secondary: Total days that a primary caretaker had to stay home from work because the infant could not attend daycare because of illness Recurrence of fever (≥37.8 °C) within 5 and 10 days after initiation of treatment Number of emergency department visits within 10 days of enrolment Hepatic and renal function Appearance of gastrointestinal symptoms or bleeding
Notes	Location: Central Israel Setting: multi‐Centre; three primary paediatric community ambulatory centers. Funding: None disclosed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Computerized random number generator to stratify according to the center in blocks of 60 numbers so that each block comprised 20 patients randomly assigned to each treatment group, with 10 patients assigned to each loading medication."
Allocation concealment (selection bias)	Low risk	Admitting nurse used a computerized random‐number generator and handed the parent or guardian a sealed opaque folder holding 3 sealed envelopes: 1 containing an advice sheet explaining the physiology of fever and its nonpharmacologic management; 1 containing the prescription for the loading medication; and 1 containing the drug prescription.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Parents were described as being blinded, but the differences in drug regimens and lack of placebos in the single agent arms suggest that blinding is unlikely.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"All of the children were evaluated and followed up by the same physician (E.M.S.), who was blinded to the group allocations (as were the parents or guardians)."
Incomplete outcome data (attrition bias) All outcomes	Low risk	A total of 480 infants met the eligibility criteria, of whom 464 (96.7%) completed the study. Of the 16 infants (3.3%) who withdrew from the study, 7 (1.5%) failed to return for follow‐up visits within the first 10 days, and 9 (1.9%) did not return for laboratory evaluation after symptoms were alleviated.
Selective reporting (reporting bias)	Unclear risk	No significant differences found with different loading medications, so patients were grouped according to maintenance medication. Data for outcomes from different loading medication groups were not reported.
Other bias	Low risk