Shah 2008.

Methods	Single‐centre randomised controlled trial performed in Canada.
Participants	110 full‐term neonates, ≥ 37 weeks' gestation, neonates who were ≥ 2500 g birth weight undergoing routine post‐natal intramuscular injection of vitamin K 0.5 mL. Exclusion criteria: neonates requiring admission to the NICU, with major congenital or know neurological abnormalities (antenatal diagnosis) or required evaluation for sepsis at birth.
Interventions	Infants received amethocaine gel 4% (55 infants) or placebo (55 infants) to the skin and covered with an adhesive Tegaderm occlusive dressing for 30 minutes before the scheduled time of the procedure.
Outcomes	Pain: measured using the facial grimacing score. The presence or absence of 3 facial actions (brow bulge, eyes squeeze and deepening of the naso‐labial furrow) were scored in 2‐second intervals for the first 20 seconds (or less if the phase lasted < 20 seconds) of each procedure phase from the videotapes by a trained research assistant who was not involved in the video recording. The research assistant scored each phase sequentially and was unaware of group assignment and study hypothesis. The data were then collapsed for each facial action into the percentage of time the infant expressed the action. An overall pain score was computed by summing the percentage scores for the 3 facial actions and then dividing by 3. The score ranged from 0% to 100%. These 3 facial actions were chosen for the study as they have been observed in 99% of neonates within 6 seconds of an invasive procedure (heel lance) and are believed to be the most sensitive indicators of infant pain. Local adverse events.
Notes	Sample size based on the pain scores obtained from a previous study. To achieve a clinically significant reduction in pain scores of 20% between groups, with an SD of 7.6, with 80% power and alpha value of < 0.05, authors estimated a sample size of 49 infants in each group.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Using a computer‐generated list created by a statistician, neonates were consecutively randomised.
Allocation concealment (selection bias)	Low risk	"Allocation concealment was achieved by keeping the randomisation code in the pharmacy department." Once an eligible neonate was identified, the research pharmacist dispensed the single‐dose study medication. Each participant was identified by a number.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Blinding parents, carers (nurses), and the research team.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"Outcome assessors were blinded to treatment allocation throughout the study. The code was not broken until the database was locked."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Outcomes for all infants reported.
Selective reporting (reporting bias)	Unclear risk	Protocol not obtained.
Other bias	Low risk