Bullard 2004.

Methods	Study design: RCT Data collection: data collected through automated tracking of database logins Unit of analysis issues: no
Participants	Participants: 10 full‐time physicians (> 75% of study site ER physicians expressed an interest of being involved in the study) Total number randomized: each of 10 volunteer ER physicians was randomized using a matched‐pair design to work 5 shifts in using standard methods (DC access) and 5 shifts with a wirelessly networked MC Practitioners lost to follow‐up: 0 Baseline characteristics of participants: Age, median (IQR): 35 years (32‐37) Gender: 100% men Qualifications: 6 (60%) had certification in emergency medicine from the College of Family Physicians of Canada, and 4 (40%) had specialty (American Board of Emergency Medicine or Fellow of the Royal College of Physicians of Canada) training (> 4 years of program training) Setting: ER (academic, tertiary‐care ER with 75,000 annual visits); non‐critical ER areas Country: Canada
Interventions	Description: MC vs. DC access to electronic CPG and other health information and a 1‐to‐1 session on use of the MC Type of intervention: Organisational: provision of MC access at point of care Study period: 24 June 2002 to 30 September 2002 Duration of intervention: 5 shifts per physician Control: 5 shifts with the standard DC
Outcomes	Number of logins to the system (utilization of electronic CPGs and other material available via the electronic system)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Each of ten volunteer emergency physicians was randomised using a matched‐pair design to work five shifts in standard fashion (desktop computer) and five shifts with a wireless networked (mobile computer)"
Allocation concealment (selection bias)	Low risk	"Concealed, block randomisation was used to allocate the work mode [mobile or desktop computer] for each physician's ten assigned shifts"
Blinding (performance bias and detection bias) All outcomes	High risk	Participants were not blinded to the intervention, DC or MC Use data for each physician/participant were compared between shifts (while assigned to the MC vs. DC)
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data were collected automatically at each login (p. 1188, col 2, para 2)
Selective reporting (reporting bias)	Low risk	Use of electronic information sources were tracked automatically by the system. "Utilizations of ...the CPGs.. for each individual were compared between shifts (while assigned to the mobile and desktop computers)" (p. 1188, col 2, para 3)