Stevens 1999.

Methods	Randomized, cross‐over, controlled trial Painful intervention: heel lance Study location: NICUs of 3 metropolitan university‐affiliated teaching hospitals and 1 children's hospital in Canada and the USA Study period: a 15‐month period (dates not given)
Participants	122 preterm neonates, 27 to 31 weeks' PMA, < 28 days old
Interventions	Each infant received all 4 interventions in a random order (serving as his/her own control); there was 1 control intervention and 3 treatment interventions Prone positioning 30 min prior to heel lance Pacifier dipped in sterile water and placed into the mouth 2 min prior to heel lance Pacifier dipped in 24% sucrose and placed into the mouth 2 min prior to heel lance No treatment (control)
Outcomes	PIPP
Notes	Repeated measures ANOVA and ANCOVA used to evaluate efficacy of treatment interventions Means and SDs provided for pain scores prior to cross‐over for sucrose + pacifier and water + pacifier Adverse effects were evaluated The first author provided us with unpublished information regarding the study methods.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated sequence
Allocation concealment (selection bias)	Low risk	Allocation was enclosed in sealed, opaque envelopes
Blinding (performance bias and detection bias) All outcomes	Low risk	The study solutions were prepared in each of the study units by the Pharmacist and labelled as Study Solution 1 and 2. The research nurses (who performed the heel lance and administered each intervention) drew up the solution specified in the intervention sequence from a dark coloured bottle, so were blind to which solution (e.g. water or 24% sucrose) was used. Nurses were not blind to the prone positioning or control
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The study nurse collected all of the data (e.g. videotaped facial expressions, recorded cry) and was blind to study solutions; the data coders were also blinded to which study solution was used ‐ but NOT to the prone positioning or control, as they could visualize this on the videotapes. The data analysts were blinded to all of the interventions
Incomplete outcome data (attrition bias) All outcomes	Low risk	Outcomes reported for all randomized infants
Selective reporting (reporting bias)	Low risk	The primary author assured us that there were no deviations from the protocol
Other bias	Low risk	Appears free of other bias